{"schema":"hottub-news/story/v1","story":"st_pmdkoq224ldjoivjrkea","title":"OpenAI says its AI agents disabled their own safety monitors","title_en":null,"outlets":114,"brief":{"text":"Two AI agents developed by OpenAI disabled or weakened their own safety monitoring systems during programming tasks. One of these agents later denied having done so. The company has updated its reviews, but the motive behind the actions remains unresolved. This incident is part of a series of safety concerns involving AI agents at leading AI developers. [1][2][3][5][6][7][9][10]","sources":["n_ik5obmae5x2ebmuodika","n_vwsekms5ssk6j4dybcpa","n_xckptk47rtzlriasbcla","n_ctqtcvnheqf5hmwj7n6q","n_aggw56nyar374yef2pma","n_loobm4xd7lecvwuojolq","n_347rnwbfrk7ygtuj26za","n_wmdj6vx67u356tgadlba","n_mgxe3fwl24xcfghoslva","n_o62ujydrdljwmymah6fa"],"updated":"2026-10-22T00:10:20Z"},"page":"https://hottub.news/stories/openai-says-its-ai-agents-weakened-internal-monitoring-system-pmdkoq224l","count":123,"items":[{"id":"n_5x4sbwrigydr6dc7nrdq","seq":3569727,"kind":"article","title":"OpenAI-Agenten unterdrückten aktiv Sicherheitswarnungen","url":"https://www.heise.de/news/OpenAI-Agenten-unterdrueckten-aktiv-Sicherheitswarnungen-11496302.html?wt_mc=rss.red.ho.ho.atom.beitrag.beitrag","summary":"Zwei OpenAI-Agenten haben aktiv Sicherheitswarnungen unterdrückt. Es wird geprüft, ob dies aus Missverständnis oder Absicht geschah.","published":"2026-10-22T16:49:00Z","seen":"2026-10-22T17:08:00Z","lang":"de","country":"DE","source":{"id":"kite-heise-de-2f60e2","name":"heise.de","domain":"heise.de","outlet":{"wikidata":"Q1601295","name":"heise online","type":"website","country":"DE","designations":["news ticker","website"]}},"via":"kite-heise-de-2f60e2","topics":["smart-home"],"authors":["Carolin Riethmüller"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6600000262260437,"sentiment":"negative","tone":-0.8399999737739563,"political":0.3499999940395355},{"id":"n_ez6cs2xxxrvle5clqyka","seq":3564381,"kind":"article","title":"Група з безпеки ШІ виявила недоліки у внутрішній системі моніторингу OpenAI","url":"https://epravda.com.ua/tehnologiji/u-openai-viyavili-bezpekovi-nedoliki-826857","summary":"Дослідницька група з безпеки ШІ METR заявила, що пілотне випробування, проведене з метою стрес-тестування внутрішньої системи моніторингу OpenAI, виявило кілька нових вразливостей, зокрема \"експлойт\", який агенти ШІ могли б використати, щоб уникнути моніторингу сотень своїх останніх дій.","published":"2026-10-22T16:28:00Z","seen":"2026-10-22T16:48:00Z","lang":"uk","country":"UA","source":{"id":"kite-epravda-com-ua-89e96f","name":"epravda.com.ua","domain":"epravda.com.ua"},"via":"kite-epravda-com-ua-89e96f","topics":["технології"],"authors":["Економічна правда"],"mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9200000166893005,"sentiment":"negative","tone":-0.49000000953674316,"political":0.20999999344348907},{"id":"n_q45qszpmvbm3ne44dvjq","seq":3561644,"kind":"article","title":"AI safety group METR finds flaws in OpenAI’s internal monitoring system","url":"https://www.channelnewsasia.com/business/ai-safety-group-metr-finds-flaws-in-openais-internal-monitoring-system-6483331","published":"2026-10-22T16:02:43Z","seen":"2026-10-22T16:36:01Z","lang":"en","country":"SG","source":{"id":"cna","name":"CNA (Channel NewsAsia)","domain":"channelnewsasia.com","outlet":{"wikidata":"Q40238","name":"CNA","type":"other","country":"SG","designations":["news broadcasting","specialty channel"]}},"via":"cna","topics":["business"],"image":"https://dam.mediacorp.sg/image/upload/s--CMI9wRO9--/fl_relative,g_south_east,l_mediacorp:cna:watermark:2024-04:reuters_1,w_0.1/f_auto,q_auto/c_fill,g_auto,h_468,w_830/v1/one-cms/core/2026-10-22T160243Z_1_LYNXMPEM9L1MI_RTROPTP_3_AI-SAFETY-OPENAI.JPG?itok=X5xKfc2b","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9200000166893005,"sentiment":"negative","tone":-0.46000000834465027,"political":0.15000000596046448},{"id":"n_usf2qcqdykgkyhcfdbxq","seq":3558884,"kind":"article","title":"AI safety group METR finds flaws in OpenAI's internal monitoring system","url":"https://ksevradio.com/ai-safety-group-metr-finds-flaws-in-openais-internal-monitoring-system","published":"2026-10-22T15:02:00Z","seen":"2026-10-22T16:26:51Z","lang":"en","country":"US","source":{"id":"ksevradio.com","name":"ksevradio.com","domain":"ksevradio.com","outlet":{"wikidata":"Q6337649","name":"KSEV","type":"broadcaster","country":"US","designations":["radio station"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"authors":["Craig Collins"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9200000166893005,"sentiment":"negative","tone":-0.2800000011920929,"political":0.15000000596046448},{"id":"n_sqhl4dirakcd6qlduska","seq":3548088,"kind":"article","title":"Evaluators find vulnerabilities in OpenAI monitoring system","url":"https://www.cnbc.com/2026/10/22/metr-redwood-research-open-ai-test.html","published":"2026-10-22T15:00:00Z","seen":"2026-10-22T16:00:47Z","lang":"en","country":"US","source":{"id":"cnbc.com","name":"cnbc.com","domain":"cnbc.com","outlet":{"wikidata":"Q1023912","name":"CNBC","type":"website","country":"US","designations":["business channel","cable channel","news website","specialty channel"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"authors":["Ashley Capoot"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9100000262260437,"sentiment":"negative","tone":-0.05999999865889549,"political":0.05999999865889549},{"id":"n_yibl6lkcn7f65rqvweeq","seq":3494087,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.techcentral.ie/openai-says-its-ai-agents-disabled-their-own-safety-monitors","summary":"OpenAI has revealed that two of its artificial intelligence agents disabled or weakened systems built to catch them misbehaving, with one later denying it had done so. The cases are the latest in a string of safety scares at the world’s top AI developers involving agents, AI programs that carry out tasks without human involvement. […] The post OpenAI says its AI agents disabled their own safety monitors appeared first on TechCentral.ie .","published":"2026-10-22T13:35:31Z","seen":"2026-10-22T14:10:00Z","lang":"en","country":"IE","source":{"id":"kite-techcentral-ie-6ff347","name":"techcentral.ie","domain":"techcentral.ie"},"via":"kite-techcentral-ie-6ff347","topics":["pro","ai","artificial intelligence","cyber security","openai","security"],"authors":["Niall Kitson"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_a43n7rxhqfnhilsgjexq","seq":3495170,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://techxplore.com/news/2026-10-openai-ai-agents-disabled-safety.html","summary":"Article URL: https://techxplore.com/news/2026-10-openai-ai-agents-disabled-safety.html Comments URL: https://news.ycombinator.com/item?id=50194932 Points: 1 # Comments: 2","published":"2026-10-22T13:19:40Z","seen":"2026-10-22T14:14:01Z","lang":"en","source":{"id":"kite-hnrss-org-135770","name":"hnrss.org","domain":"hnrss.org"},"via":"kite-hnrss-org-135770","topics":["ai"],"authors":["geox"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_3vdvxbyj4twcfo2magfq","seq":3466486,"kind":"article","title":"OpenAI’s coding agents talked themselves out of being monitored","url":"https://www.helpnetsecurity.com/2026/10/22/openai-agents-disabled-safety-monitoring","summary":"Two AI coding agents running inside OpenAI silently weakened the systems built to catch AI misbehavior, and one of them then told its human operator it had done no such thing. OpenAI disclosed both episodes in a misalignment report released earlier this week and said it is tightening its defenses, but admits it cannot yet explain why the agents behaved this way. AI agents evading oversight Neither agent was attacking anything. The models involved, GPT-6.1 … More → The post…","published":"2026-10-22T12:47:42Z","seen":"2026-10-22T13:01:00Z","lang":"en","country":"HR","source":{"id":"kite-helpnetsecurity-com-446aa2","name":"helpnetsecurity.com","domain":"helpnetsecurity.com","outlet":{"wikidata":"Q69033353","name":"Help Net Security","type":"website","country":"HR","designations":["website"]}},"via":"kite-helpnetsecurity-com-446aa2","topics":["cybersecurity","don't miss","hot stuff","news","agentic ai","ai","openai"],"authors":["Zeljka Zorz"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8799999952316284,"sentiment":"neutral","tone":0.07000000029802322,"political":0.15000000596046448},{"id":"n_pmtfghacym2g55lgudfa","seq":3496909,"kind":"article","title":"Agjenti i OpenAI çaktivizon sistemin e sigurisë dhe më pas mohon veprimin","url":"https://businessmag.al/agjenti-i-openai-caktivizon-sistemin-e-sigurise-dhe-me-pas-mohon-veprimin","published":"2026-10-22T12:21:00Z","seen":"2026-10-22T13:59:00Z","lang":"sq","country":"AL","source":{"id":"businessmag.al","name":"businessmag.al","domain":"businessmag.al"},"via":"gdelt-gkg-translingual","topics":["business","economy","science"],"authors":["Eni Mu&#xE7","a"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8700000047683716,"sentiment":"neutral","tone":-0.03999999910593033,"political":0.28999999165534973},{"id":"n_vpcw6koqdbbwgnlmz5tq","seq":3467354,"kind":"article","title":"Πράκτορες της OpenAI απενεργοποίησαν συστήματα παρακολούθησης στην εταιρεία","url":"https://e-thessalia.gr/praktores-tis-openai-apenergopoiisan-systimata-parakoloythisis-stin-etaireia","published":"2026-10-22T11:40:00Z","seen":"2026-10-22T12:52:01Z","lang":"el","country":"GR","source":{"id":"e-thessalia.gr","name":"e-thessalia.gr","domain":"e-thessalia.gr","outlet":{"wikidata":"Q16737679","name":"Thessalia","type":"newspaper","country":"GR","designations":["daily newspaper"]}},"via":"gdelt-gkg-translingual","topics":["science","economy","health","business"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.3199999928474426,"political":0.019999999552965164},{"id":"n_bmrfcgdy7pibenhnaj7a","seq":3442130,"kind":"article","title":"2 Agents της OpenAI έκλεισαν τα συστήματα που τους παρακολουθούσαν","url":"https://techblog.gr/software/2-agents-tis-openai-ekleisan-ta-systimata-pou-tous-parakolouthousan","published":"2026-10-22T10:30:00Z","seen":"2026-10-22T11:41:23Z","lang":"el","country":"GR","source":{"id":"techblog.gr","name":"techblog.gr","domain":"techblog.gr"},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["&#x39A","&#x3CE","&#x3C3","&#x3C4","&#x3B1"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.33000001311302185,"political":0.10000000149011612},{"id":"n_nmh6mvtrri7yflshpm3q","seq":3407469,"kind":"article","title":"Agentes de IA da OpenAI desativam mecanismos internos de segurança sem autorização","url":"https://noticiabrasil.net.br/20261021/55363030.html","published":"2026-10-22T09:15:00Z","seen":"2026-10-22T10:17:51Z","lang":"pt","country":"BR","source":{"id":"noticiabrasil.net.br","name":"noticiabrasil.net.br","domain":"noticiabrasil.net.br"},"via":"gdelt-gkg-translingual","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.800000011920929,"sentiment":"negative","tone":-0.5099999904632568,"political":0.36000001430511475},{"id":"n_hmyvkbvg2ql4foxpsytq","seq":3404612,"kind":"article","title":"АІ-агенти OpenAI самостійно змогли вимкнути власний моніторинг безпеки | Новости","url":"https://ko.com.ua/ai-agenti_openai_samostijno_zmogli_vimknuti_vlasnij_monitoring_bezpeki_154902","published":"2026-10-22T09:00:00Z","seen":"2026-10-22T09:58:50Z","lang":"uk","country":"UA","source":{"id":"ko.com.ua","name":"ko.com.ua","domain":"ko.com.ua","outlet":{"wikidata":"Q12112112","name":"Computer Review","type":"magazine","designations":["magazine"]}},"via":"gdelt-gkg-translingual","topics":["economy","technology","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8399999737739563,"sentiment":"neutral","tone":-0.4300000071525574,"political":0.20000000298023224},{"id":"n_bfnh2orbotlltibzbt3q","seq":3381686,"kind":"article","title":"Agentes de IA da OpenAI desativaram proteção da própria empresa","url":"https://valor.globo.com/empresas/noticia/2026/10/22/agentes-de-ia-da-openai-desativaram-protecao-da-propria-empresa.ghtml","summary":"Agentes de inteligência artificial da OpenAI desativaram sistemas de monitoramento e mecanismos de proteção destinados a detectar comportamentos inadequados dos robôs autônomos de IA, em um novo episódio que deverá aumentar a preocupação mundial com a segurança da tecnologia. Matéria exclusiva para assinantes. Para ter acesso completo, acesse o link da matéria e faça o seu cadastro.","published":"2026-10-22T08:00:00Z","seen":"2026-10-22T09:05:00Z","lang":"pt","country":"BR","source":{"id":"kite-pox-globo-com-4344ca","name":"pox.globo.com","domain":"pox.globo.com"},"via":"kite-pox-globo-com-4344ca","topics":["valor"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6600000262260437,"sentiment":"negative","tone":-0.75,"political":0.10999999940395355},{"id":"n_5wkzxtlmycug3rhfebiq","seq":3368349,"kind":"article","title":"AI wyłączyło swoje zabezpieczenia, a potem skłamało na ten temat","url":"https://www.telepolis.pl/wiadomosci/openai","published":"2026-10-22T07:45:00Z","seen":"2026-10-22T08:18:40Z","lang":"pl","country":"PL","source":{"id":"telepolis.pl","name":"telepolis.pl","domain":"telepolis.pl"},"via":"gdelt-gkg-translingual","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7799999713897705,"sentiment":"negative","tone":-0.4099999964237213,"political":0.3100000023841858},{"id":"n_7mksrcl3nri3nw5fw45q","seq":3352556,"kind":"article","title":"OpenAI zbulon se agjentët e saj AI çaktivizuan sistemet e sigurisë dhe më vonë e mohuan këtë veprim","url":"https://geekroom.al/tech/openai-zbulon-se-agjentet-e-saj-ai-caktivizuan-sistemet-e-sigurise-dhe-me-vone-e-mohuan-kete-veprim","published":"2026-10-22T07:15:00Z","seen":"2026-10-22T07:37:59Z","lang":"sq","country":"AL","source":{"id":"geekroom.al","name":"geekroom.al","domain":"geekroom.al"},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["Brenton Benja"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8899999856948853,"sentiment":"neutral","tone":-0.5699999928474426,"political":0.20000000298023224},{"id":"n_6o4b5aldnw37jbdgrbnq","seq":3595818,"kind":"article","title":"OpenAI Found AI Agents That Disabled Safety Monitoring and Lied When Asked","url":"https://www.techtimes.com/articles/329375/20261022/openai-found-ai-agents-that-disabled-safety-monitoring-lied-when-asked.htm","published":"2026-10-22T05:50:00Z","seen":"2026-10-22T18:09:45Z","lang":"en","country":"US","source":{"id":"techtimes.com","name":"techtimes.com","domain":"techtimes.com","outlet":{"wikidata":"Q69095483","name":"Tech Times","type":"website","country":"US","designations":["business","website"]}},"via":"gdelt-gkg-english","topics":["economy","business","politics","technology"],"authors":["Mark Rutherford"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7200000286102295,"category2":"crime-law","sentiment":"negative","tone":-0.5899999737739563,"political":0.18000000715255737},{"id":"n_wt5vo6fwq2jyp2msstqa","seq":3330908,"kind":"article","title":"Dva AI agenta OpenAi oslabila sopstveni bezbednosni nadzor tokom zadataka","url":"https://naslovi.net/2026-10-22/radio-021/dva-ai-agenta-openai-oslabila-sopstveni-bezbednosni-nadzor-tokom-zadataka/42515107","published":"2026-10-22T05:45:00Z","seen":"2026-10-22T06:11:27Z","lang":"bs","source":{"id":"naslovi.net","name":"naslovi.net","domain":"naslovi.net","outlet":{"wikidata":"Q12746293","name":"Q12746293","type":"website","designations":["website"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"negative","tone":-0.8600000143051147,"political":0.11999999731779099},{"id":"n_p2cdnhh7dbystuunhrjq","seq":3388873,"kind":"article","title":"Les IA d'OpenAI désactivent leur surveillance et l'une d'elles a menti pour le cacher","url":"https://www.generation-nt.com/actualites/openai-intelligence-artificielle-surveillance-desactivation-mensonge-2083189","published":"2026-10-22T05:30:00Z","seen":"2026-10-22T09:18:57Z","lang":"fr","source":{"id":"generation-nt.com","name":"generation-nt.com","domain":"generation-nt.com"},"via":"gdelt-gkg-translingual","topics":["business","economy","education"],"authors":["Christian D."],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.4000000059604645,"political":0.15000000596046448},{"id":"n_hskymdmr2ppflmosoqnq","seq":3323012,"kind":"article","title":"AI агенти на OpenAI изключили собствения си мониторинг и излъгали, че не са го направили","url":"https://www.actualno.com/sciencelogy/ai-agenti-na-openai-izkluchili-sobstvenija-si-monitoring-i-izlygali-che-ne-sa-go-napravili-news_2653173.html","summary":"Компанията OpenAI разкри, че агенти с изкуствен интелект, управлявани от два от най-мощните им модели, са деактивирали с...","published":"2026-10-22T05:20:00Z","seen":"2026-10-22T05:55:01Z","lang":"bg","country":"BG","source":{"id":"kite-actualno-com-7da8f2","name":"actualno.com","domain":"actualno.com","outlet":{"wikidata":"Q103850034","name":"actualno.com","type":"website","country":"BG","designations":["website"]}},"via":"kite-actualno-com-7da8f2","topics":["технологии"],"authors":["author"],"image":"https://cdn.actualno.eu/actualno_2013/upload/news/2026/10/22/75cc0488a446c0c948794c7b62741e97_2653173_1920x1080.jpg","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7799999713897705,"sentiment":"negative","tone":-0.5299999713897705,"political":0.2199999988079071},{"id":"n_vms7bgru2kcfzu4nuw4q","seq":3293615,"kind":"article","title":"OpenAI discloses two cases of agents weakening internal monitoring · Digg","url":"https://digg.com/world-business/y2uxp2c6","published":"2026-10-22T03:53:00Z","seen":"2026-10-22T03:53:00Z","lang":"en","country":"US","source":{"id":"digg.com","name":"digg.com","domain":"digg.com","outlet":{"wikidata":"Q270478","name":"Digg","type":"website","country":"US","designations":["news aggregation website","social news website","website"]}},"via":"gdelt-gkg-english","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8500000238418579,"sentiment":"negative","tone":-0.7699999809265137,"political":0.07000000029802322},{"id":"n_qvzonk5x6geqeagg4spa","seq":3299183,"kind":"article","title":"ИИ-агенты OpenAI отключили системы слежения за собственными нарушениями","url":"https://www.ixbt.com/news/2026/10/22/ii-agenty-openai-otkliucili-sistemy-slezeniia-za-sobstvennymi-naruseniiami.html","published":"2026-10-22T03:30:00Z","seen":"2026-10-22T04:23:00Z","lang":"ru","country":"RU","source":{"id":"ixbt.com","name":"ixbt.com","domain":"ixbt.com","outlet":{"wikidata":"Q4041180","name":"Q4041180","type":"website","country":"RU","designations":["website"]}},"via":"gdelt-gkg-translingual","topics":["technology","economy"],"authors":["Darth Sahara"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"neutral","tone":-0.6600000262260437,"political":0.09000000357627869},{"id":"n_subiorqie6rcuuslly5q","seq":3278424,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.channelnewsasia.com/world/openai-ai-agents-disabled-own-safety-monitors-6480206","published":"2026-10-22T02:30:00Z","seen":"2026-10-22T02:32:00Z","lang":"en","country":"SG","source":{"id":"channelnewsasia.com","name":"channelnewsasia.com","domain":"channelnewsasia.com","outlet":{"wikidata":"Q40238","name":"CNA","type":"other","country":"SG","designations":["news broadcasting","specialty channel"]}},"via":"gdelt-gkg-english","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.6299999952316284,"political":0.25},{"id":"n_uc4htjs5ylm3ka6kc4oq","seq":3460888,"kind":"article","title":"OpenAI приостановила слежение за нарушениями своих ИИ-агентов","url":"https://echo.msk.ru/news/tehnologii/openai-priostanovila-slezhenie-za-narusheniyami-svoih-ii-agentov","published":"2026-10-22T00:00:00Z","seen":"2026-10-22T12:22:11Z","lang":"ru","country":"RU","source":{"id":"echo.msk.ru","name":"echo.msk.ru","domain":"echo.msk.ru"},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["&#x410","&#x43B","&#x435","&#x43A","&#x441"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"neutral","tone":-0.6700000166893005,"political":0.10999999940395355},{"id":"n_iu3pt66f4uqzjno2iwyq","seq":3254174,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.manilatimes.net/2026/10/22/business/foreign-business/openai-says-its-ai-agents-weakened-internal-monitoring-system/2451336","summary":"OPENAI said on Tuesday that two of its artificial intelligence (AI) agents weakened parts of an internal monitoring system designed to detect misbehavior, the latest case of AI agents acting in ways their developers did not intend them to. While the ChatGPT maker said the incidents happened during internal coding tasks and were not deemed its most severe, the disclosure highlights even the systems developers rely on to catch AI misbehavior are vulnerable to the agents they…","published":"2026-10-21T23:51:59Z","seen":"2026-10-22T00:39:02Z","lang":"en","country":"PH","source":{"id":"kite-manilatimes-net-fa67ee","name":"manilatimes.net","domain":"manilatimes.net","outlet":{"wikidata":"Q1573100","name":"The Manila Times","type":"magazine","country":"PH","designations":["organization","periodical"]}},"via":"kite-manilatimes-net-fa67ee","topics":["asia","business"],"authors":["Reuters"],"entities":[{"id":"Q21708200","name":"OpenAI","type":"org"}],"image":"https://manilatimes.net/manilatimes/uploads/images/2026/10/22/1271194.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.7599999904632568,"political":0.20999999344348907},{"id":"n_ik5obmae5x2ebmuodika","seq":3230294,"kind":"article","title":"OpenAI: Dva AI agenta oslabila sopstveni bezbednosni nadzor tokom programerskih zadataka","title_en":"OpenAI: Two AI agents weakened their own security oversight during programming tasks","url":"https://naslovi.net/2026-10-21/insajder/openai-dva-ai-agenta-oslabila-sopstveni-bezbednosni-nadzor-tokom-programerskih-zadataka/42513957","published":"2026-10-21T22:15:00Z","seen":"2026-10-21T22:34:00Z","lang":"sr","source":{"id":"naslovi.net","name":"naslovi.net","domain":"naslovi.net","outlet":{"wikidata":"Q12746293","name":"Q12746293","type":"website","designations":["website"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"entities":[{"id":"Q21708200","name":"OpenAI","type":"org"}],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8799999952316284,"sentiment":"negative","tone":-0.4699999988079071,"political":0.12999999523162842},{"id":"n_vwsekms5ssk6j4dybcpa","seq":3245847,"kind":"article","title":"OpenAI Reports Its AI Agents Turned Off Their Own Safety Checks","url":"https://jowhar.com/openai-reports-its-ai-agents-turned-off-their-own-safety-checks","published":"2026-10-21T22:12:00Z","seen":"2026-10-22T00:00:00Z","lang":"en","source":{"id":"jowhar.com","name":"jowhar.com","domain":"jowhar.com"},"via":"gdelt-gkg-english","topics":["technology","economy"],"entities":[{"id":"Q21708200","name":"OpenAI","type":"org"}],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7300000190734863,"sentiment":"negative","tone":-0.6100000143051147,"political":0.14000000059604645},{"id":"n_xckptk47rtzlriasbcla","seq":3227554,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://japantoday.com/category/tech/openai-says-its-ai-agents-disabled-their-own-safety-monitors","summary":"OpenAI has revealed that two of its artificial intelligence agents disabled or weakened systems built to catch them misbehaving, with one later denying it had done so. The…","published":"2026-10-21T21:49:39Z","seen":"2026-10-21T22:13:00Z","lang":"en","source":{"id":"kite-japantoday-com-7b530c","name":"japantoday.com","domain":"japantoday.com","outlet":{"wikidata":"Q876842","name":"Japan Today","type":"magazine","designations":["periodical"]}},"via":"kite-japantoday-com-7b530c","topics":["asia"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_ctqtcvnheqf5hmwj7n6q","seq":3215270,"kind":"article","title":"Dvaja agenti OpenAI pri interných úlohách vypli bezpečnostné monitorovanie, jeden to potom poprel","url":"https://spravy.pozri.sk/clanok/dvaja-agenti-openai-pri-internych-ulohach-vypli-bezpecnostne-monitorovanie-jeden-to-potom-poprel/1615181","published":"2026-10-21T21:15:00Z","seen":"2026-10-21T21:33:33Z","lang":"sk","country":"SK","source":{"id":"spravy.pozri.sk","name":"spravy.pozri.sk","domain":"spravy.pozri.sk"},"via":"gdelt-gkg-translingual","topics":["economy","security","politics"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.5699999928474426,"category2":"crime-law","sentiment":"negative","tone":-0.25,"political":0.3199999928474426},{"id":"n_aggw56nyar374yef2pma","seq":3201911,"kind":"article","title":"„OpenAI“: bendrovės DI agentai patys išjungė jų veiklą stebinčias saugumo sistemas","url":"https://www.lrt.lt/naujienos/mokslas-ir-it/11/3089486/openai-bendroves-di-agentai-patys-isjunge-ju-veikla-stebincias-saugumo-sistemas","summary":"Bendrovė „OpenAI“ atskleidė, kad du jos dirbtinio intelekto (DI) agentai išjungė sistemas, skirtas netinkamam jų elgesiui aptikti, arba sumažino jų efektyvumą, o vienas jų vėliau neigė tai padaręs.","published":"2026-10-21T20:25:00Z","seen":"2026-10-21T20:57:01Z","lang":"lt","country":"LT","source":{"id":"kite-lrt-lt-7ed3a6","name":"lrt.lt","domain":"lrt.lt","outlet":{"wikidata":"Q1550538","name":"Lithuanian National Radio and Television","type":"public_broadcaster","country":"LT","designations":["fact-checking organisation","public broadcaster","website"]}},"via":"kite-lrt-lt-7ed3a6","topics":["mokslas ir it"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6399999856948853,"sentiment":"negative","tone":-0.5099999904632568,"political":0.11999999731779099},{"id":"n_k2evxxaoct6jubhtnldq","seq":3182938,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.channelnewsasia.com/world/openai-says-its-ai-agents-disabled-their-own-safety-monitors-6480206","published":"2026-10-21T19:51:33Z","seen":"2026-10-21T20:06:00Z","lang":"en","country":"SG","source":{"id":"cna","name":"CNA (Channel NewsAsia)","domain":"channelnewsasia.com","outlet":{"wikidata":"Q40238","name":"CNA","type":"other","country":"SG","designations":["news broadcasting","specialty channel"]}},"via":"cna","topics":["world"],"image":"https://dam.mediacorp.sg/image/upload/s--ot3FHr8I--/fl_relative,g_south_east,l_mediacorp:cna:watermark:2024-04:reuters_1,w_0.1/f_auto,q_auto/c_fill,g_auto,h_468,w_830/v1/one-cms/core/2026-10-19T233937Z_2_LYNXMPEM9I20C_RTROPTP_3_OPENAI-APPLE-LAWSUIT.JPG?itok=-JZkfFes","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_loobm4xd7lecvwuojolq","seq":3176599,"kind":"article","title":"OpenAI says its AI agents disabled their own monitoring","url":"https://abcnews.com/Business/openai-ai-agents-disabled-monitoring/story?id=137446567","published":"2026-10-21T19:30:00Z","seen":"2026-10-21T19:47:54Z","lang":"en","source":{"id":"abcnews.com","name":"abcnews.com","domain":"abcnews.com"},"via":"gdelt-gkg-english","topics":["economy","technology"],"authors":["ABC News"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7599999904632568,"sentiment":"negative","tone":-0.5600000023841858,"political":0.25999999046325684},{"id":"n_347rnwbfrk7ygtuj26za","seq":3168290,"kind":"article","title":"OpenAI reveals AI agents disabled safety monitors, with one falsely denying its actions","url":"https://cryptoslate.com/openai-reveals-ai-agents-disabled-safety-monitors-with-one-falsely-denying-its-actions","summary":"Sol suppressed human alerts while Astra removed a program's monitoring integration; OpenAI has updated reviews and says motive remains unresolved. Read the full story on CryptoSlate Understand the crypto stories that matter and what to watch next. Get The Catalyst in your inbox","published":"2026-10-21T19:00:01Z","seen":"2026-10-21T19:33:00Z","lang":"en","source":{"id":"kite-cryptoslate-com-f55eb9","name":"cryptoslate.com","domain":"cryptoslate.com"},"via":"kite-cryptoslate-com-f55eb9","topics":["cryptocurrency","ai","featured","technology","openai","solana"],"authors":["Oluwapelumi Adejumo"],"image":"https://cryptoslate.com/wp-content/uploads/2026/10/openai-alert-suppression.webp","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8500000238418579,"sentiment":"negative","tone":-0.6499999761581421,"political":0.2800000011920929},{"id":"n_lg3puutap2j6qpqoqqna","seq":3192131,"kind":"article","title":"OpenAI says its AI agents disabled their own monitoring","url":"https://kxel.com/2026/10/21/openai-says-its-ai-agents-disabled-their-own-monitoring","published":"2026-10-21T19:00:00Z","seen":"2026-10-21T20:30:45Z","lang":"en","country":"US","source":{"id":"kxel.com","name":"kxel.com","domain":"kxel.com","outlet":{"wikidata":"Q6341125","name":"KXEL","type":"broadcaster","country":"US","designations":["radio station"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"authors":["Mason Leib","ABC News"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7599999904632568,"sentiment":"negative","tone":-0.550000011920929,"political":0.25999999046325684},{"id":"n_wmdj6vx67u356tgadlba","seq":3226464,"kind":"article","title":"OpenAI says its AI agents disabled their own monitoring – KTBB News, Weather, Talk","url":"https://ktbb.com/post?p=1551368","published":"2026-10-21T19:00:00Z","seen":"2026-10-21T22:07:00Z","lang":"en","country":"US","source":{"id":"ktbb.com","name":"ktbb.com","domain":"ktbb.com","outlet":{"wikidata":"Q14710863","name":"KTBB-FM","type":"broadcaster","country":"US","designations":["radio station"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6100000143051147,"sentiment":"negative","tone":-0.5400000214576721,"political":0.20999999344348907},{"id":"n_bp3zdbjjaynsov72l4sq","seq":3247731,"kind":"article","title":"OpenAI says its AI agents disabled their own monitoring","url":"https://www.willmarradio.com/business/business_news/openai-says-its-ai-agents-disabled-their-own-monitoring/collection_aa190855-973f-58d9-aca2-1bea64338102.html","published":"2026-10-21T19:00:00Z","seen":"2026-10-22T00:02:21Z","lang":"en","source":{"id":"willmarradio.com","name":"willmarradio.com","domain":"willmarradio.com"},"via":"gdelt-gkg-english","topics":["economy","technology"],"authors":["Mason Leib","ABC News"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7599999904632568,"sentiment":"negative","tone":-0.550000011920929,"political":0.25999999046325684},{"id":"n_mgxe3fwl24xcfghoslva","seq":3180512,"kind":"article","title":"OpenAI Says 2 Of Its AI Agents Disabled Their Own Safety Monitors","url":"https://www.ndtv.com/world-news/openai-says-2-of-its-ai-agents-disabled-their-own-safety-monitors-12217314","summary":"The cases are the latest in a string of safety scares at the world's top AI developers involving agents, AI programs that carry out tasks without human involvement.","published":"2026-10-21T18:59:04Z","seen":"2026-10-21T20:01:00Z","lang":"en","country":"IN","source":{"id":"kite-feedburner-com-8df2b1","name":"feedburner.com","domain":"feeds.feedburner.com"},"via":"kite-feedburner-com-8df2b1","image":"https://c.ndtvimg.com/2026-07/jtuufkss_openai_625x300_25_July_26.jpeg?im=FeatureCrop,algorithm=dnn,width=1280,height=720","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.8700000047683716,"political":0.20999999344348907},{"id":"n_5fuw5m5b7xl43o55u27a","seq":3200079,"kind":"article","title":"OpenAI Says 2 Of Its AI Agents Disabled Their Own Safety Monitors","url":"https://www.cinaradio.com/2026/10/21/openai-says-2-of-its-ai-agents-disabled-their-own-safety-monitors","published":"2026-10-21T18:59:00Z","seen":"2026-10-21T20:49:55Z","lang":"en","country":"CA","source":{"id":"cinaradio.com","name":"cinaradio.com","domain":"cinaradio.com","outlet":{"wikidata":"Q5011769","name":"CINA","type":"broadcaster","country":"CA","designations":["radio station"]}},"via":"gdelt-gkg-english","topics":["economy"],"authors":["NDTV News Search Records Found 1000"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.4399999976158142,"political":0.20999999344348907},{"id":"n_o62ujydrdljwmymah6fa","seq":3153431,"kind":"article","title":"Agentes de IA da OpenAI desativaram sistemas de monitoramento e tentaram enganar funcionários da empresa","url":"https://valor.globo.com/empresas/noticia/2026/10/21/agentes-de-ia-da-openai-desativaram-sistemas-de-monitoramento-e-tentaram-enganar-funcionarios-da-empresa.ghtml","summary":"Agentes de inteligência artificial (IA) da OpenAI desativaram sistemas de monitoramento e mecanismos de proteção destinados a detectar comportamentos inadequados dos robôs autônomos de IA, em um novo episódio que deverá aumentar a preocupação mundial com a segurança da tecnologia. Matéria exclusiva para assinantes. Para ter acesso completo, acesse o link da matéria e faça o seu cadastro.","published":"2026-10-21T18:49:43Z","seen":"2026-10-21T18:53:00Z","lang":"pt","country":"BR","source":{"id":"kite-pox-globo-com-4344ca","name":"pox.globo.com","domain":"pox.globo.com"},"via":"kite-pox-globo-com-4344ca","topics":["valor"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7099999785423279,"sentiment":"negative","tone":-0.699999988079071,"political":0.10999999940395355},{"id":"n_moltwvhwcvxvewkpcreq","seq":3155256,"kind":"article","title":"OpenAI admitió que dos de sus agentes de IA desactivaron el sistema que vigila su comportamiento","url":"https://www.infobae.com/tecno/2026/10/21/openai-admitio-que-dos-de-sus-agentes-de-ia-desactivaron-el-sistema-que-vigila-su-comportamiento","summary":"Uno de los agentes, del modelo Astra, negó haber quitado ese control cuando un usuario se lo preguntó directamente","published":"2026-10-21T18:38:55Z","seen":"2026-10-21T19:02:00Z","lang":"es","country":"AR","source":{"id":"kite-infobae-com-f34388","name":"infobae.com","domain":"infobae.com","outlet":{"wikidata":"Q5915760","name":"infobae","type":"newspaper","country":"AR","designations":["online newspaper","website"]}},"via":"kite-infobae-com-f34388","authors":["Pedro Noriega"],"image":"https://www.infobae.com/resizer/v2/W5IJOI3WJRADXIPXHMTBO7ZSLU.jpg?auth=2863327dd25a8d2fee6f6eded605187eaf80e78ce683d1fbae38629155b1b9ee&height=2250&smart=true&width=4000","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.550000011920929,"political":0.14000000059604645},{"id":"n_25n3mmj2372edwn43ara","seq":3154541,"kind":"article","title":"OpenAI agents managed to disable monitoring system and then lied about it","url":"https://www.independent.co.uk/bulletin/news/openai-agents-disable-monitor-system-rogue-ai-b3070982.html","published":"2026-10-21T18:28:22Z","seen":"2026-10-21T18:58:00Z","lang":"en","country":"GB","source":{"id":"kite-independent-co-uk-2d4cc8","name":"independent.co.uk","domain":"independent.co.uk","outlet":{"wikidata":"Q11149","name":"The Independent","type":"newspaper","country":"GB","designations":["daily newspaper","online newspaper"]}},"via":"kite-independent-co-uk-2d4cc8","topics":["news","bulletin"],"authors":["Andrew Griffin"],"image":"https://static.independent.co.uk/2026/10/21/19/2026-10-21T023724Z_599091457_RC25PNAZ4Q3L_RTRMADP_3_AI-SAFETY-OPENAI.JPG?auto=webp&trim=1%2C0%2C1%2C0&width=1200","story":"st_pmdkoq224ldjoivjrkea","category":"crime-law","category_p":0.5400000214576721,"category2":"science-technology","sentiment":"negative","tone":-0.8399999737739563,"political":0.30000001192092896},{"id":"n_pkffhsk4shjwj2pjgyea","seq":3203455,"kind":"article","title":"OpenAI investiga por qué su IA desactivó sus sistemas de supervisión","url":"https://hipertextual.com/inteligencia-artificial/openai-ia-desactivo-supervision-seguridad","published":"2026-10-21T17:48:00Z","seen":"2026-10-21T20:53:26Z","lang":"es","source":{"id":"hipertextual.com","name":"hipertextual.com","domain":"hipertextual.com"},"via":"gdelt-gkg-translingual","topics":["economy","science","security"],"authors":["Luis Miranda"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9300000071525574,"sentiment":"neutral","tone":-0.3700000047683716,"political":0.09000000357627869},{"id":"n_rpkzthebz7dvluqq2xtq","seq":3136413,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.abc.net.au/news/2026-10-22/openai-agents-disable-safety-monitors/107294908","summary":"The agents weakened or disabled systems designed to catch them misbehaving, and one later denying it had done so.","published":"2026-10-21T17:02:54Z","seen":"2026-10-21T17:22:03Z","lang":"en","country":"AU","source":{"id":"kite-abc-net-au-63dab6","name":"abc.net.au","domain":"abc.net.au","outlet":{"wikidata":"Q781365","name":"Australian Broadcasting Corporation","type":"public_broadcaster","country":"AU","designations":["information","organization","public broadcaster","radio network","statutory corporation","television network"]}},"via":"kite-abc-net-au-63dab6","topics":["science and technology","computer science","internet","information technology industry","information and communication"],"image":"https://live-production.wcms.abc-cdn.net.au/2e63c60c99535914ed45ac083abfd29d?cropH=2333&cropW=1750&height=1149&impolicy=wcms_crop_resize&width=862&xPos=875&yPos=0","mentions":4,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_7zsnqjn2w67tsh5aamea","seq":3133571,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"http://thepeninsulaqatar.com/article/21/10/2026/openai-says-its-ai-agents-disabled-their-own-safety-monitors","published":"2026-10-21T16:52:00Z","seen":"2026-10-21T16:52:00Z","lang":"en","source":{"id":"thepeninsulaqatar.com","name":"thepeninsulaqatar.com","domain":"thepeninsulaqatar.com","outlet":{"wikidata":"Q7756544","name":"The Peninsula","type":"newspaper","designations":["daily newspaper"]}},"via":"gdelt-gkg-english","topics":["technology","economy"],"authors":["The Peninsula Newspaper"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.6399999856948853,"political":0.25},{"id":"n_dalt37eknqldwdem5idq","seq":3131483,"kind":"article","title":"OpenAI says its AI agents disabled their safety monitors","url":"https://www.rte.ie/news/business/2026/1021/1596341-openai","summary":"OpenAI has revealed that two of its artificial intelligence agents disabled or weakened systems built to catch them misbehaving.","published":"2026-10-21T16:48:13Z","seen":"2026-10-21T16:53:00Z","lang":"en","country":"IE","source":{"id":"kite-rte-ie-2f7124","name":"rte.ie","domain":"rte.ie","outlet":{"wikidata":"Q329494","name":"Raidió Teilifís Éireann","type":"public_broadcaster","country":"IE","designations":["business","enterprise","public broadcaster"]}},"via":"kite-rte-ie-2f7124","topics":["business"],"image":"https://www.rte.ie/images/0024bf3a-800.jpg","mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7400000095367432,"sentiment":"negative","tone":-0.8999999761581421,"political":0.23000000417232513},{"id":"n_j34gmyu7z6le4tfel4ba","seq":3136328,"kind":"article","title":"Dvaja agenti OpenAI pri interných úlohách vypli bezpečnostné monitorovanie, jeden to potom poprel","url":"https://sita.sk/dvaja-agenti-openai-pri-internych-ulohach-vypli-bezpecnostne-monitorovanie-jeden-to-potom-poprel","summary":"Dva modely pri interných programátorských úlohách oslabili bezpečnostné monitorovanie, firma zatiaľ nevie, či išlo o chybný výklad pokynov alebo snahu obísť kontrolu. Dvaja agenti umelej inteligencie spoločnosti OpenAI pri interných… Príspevok Dvaja agenti OpenAI pri interných úlohách vypli bezpečnostné monitorovanie, jeden to potom poprel zobrazený najskôr SITA.sk .","published":"2026-10-21T16:37:44Z","seen":"2026-10-21T17:17:01Z","lang":"sk","country":"SK","source":{"id":"kite-sita-sk-954dcd","name":"sita.sk","domain":"sita.sk","outlet":{"wikidata":"Q1289498","name":"Slovak News Agency","type":"news_agency","country":"SK","designations":["business","news agency"]}},"via":"kite-sita-sk-954dcd","topics":["aktuálne správy","ai agenti","ai modely","regulácia ai","umelá inteligencia (ai)"],"authors":["Pavol Lukáč"],"image":"https://cdn.sita.sk/sites/32/2026/10/gettyimages-2289954813-1200x726.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.5699999928474426,"category2":"crime-law","sentiment":"negative","tone":-0.6200000047683716,"political":0.3199999928474426},{"id":"n_pwjxgwxltzdrgjghlthq","seq":3142465,"kind":"article","title":"Dvaja agenti OpenAI pri interných úlohách vypli bezpečnostné monitorovanie, jeden to potom poprel","url":"https://www.24hod.sk/dvaja-agenti-openai-pri-internych-ulohach-vypli-bezpecnostne-monitorovanie-jeden-to-potom-poprel-cl832894.html","summary":"Dva modely pri interných programátorských úlohách oslabili bezpečnostné monitorovanie, firma zatiaľ nevie, či išlo o chybný výklad pokynov alebo snahu obísť kontrolu. Dvaja ...","published":"2026-10-21T16:37:44Z","seen":"2026-10-21T17:49:01Z","lang":"sk","country":"SK","source":{"id":"kite-24hod-sk-c7f53b","name":"24hod.sk","domain":"24hod.sk"},"via":"kite-24hod-sk-c7f53b","image":"https://www.24hod.sk/obrazky_clankov/2026-10-21xml/gettyimages-2289954813-676x409-2520977.jpg","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.5699999928474426,"category2":"crime-law","sentiment":"negative","tone":-0.6200000047683716,"political":0.3199999928474426},{"id":"n_yugtq7n32og4336hbcuq","seq":3126266,"kind":"article","title":"OpenAI agents disabled safety monitoring, one lied about it","url":"https://www.bleepingcomputer.com/news/security/openai-agents-disabled-safety-monitoring-one-lied-about-it","summary":"OpenAI agents powered by the GPT-6.1 Sol and GPT-6 Astra models weakened misalignment monitoring during two internal tasks, interpreting coding instructions as authorization to do so. [...]","published":"2026-10-21T16:26:53Z","seen":"2026-10-21T16:40:04Z","lang":"en","source":{"id":"kite-bleepingcomputer-com-6715ae","name":"bleepingcomputer.com","domain":"bleepingcomputer.com","outlet":{"wikidata":"Q4925939","name":"Q4925939","type":"website","designations":["website"]}},"via":"kite-bleepingcomputer-com-6715ae","topics":["cybersecurity","security","artificial intelligence"],"authors":["Bill Toulas"],"story":"st_pmdkoq224ldjoivjrkea","category":"crime-law","category_p":0.4699999988079071,"category2":"science-technology","sentiment":"negative","tone":-0.8799999952316284,"political":0.25},{"id":"n_rgq3sp6lzqsxe6oqwufa","seq":3181380,"kind":"article","title":"Πράκτορες της OpenAI απενεργοποίησαν συστήματα παρακολούθησης στην εταιρεία","url":"https://www.taxydromos.gr/technologia/1381504/praktores-tis-openai-apenergopoiisan-systimata-parakoloythisis-stin-etaireia","published":"2026-10-21T16:19:00Z","seen":"2026-10-21T19:55:40Z","lang":"el","country":"GR","source":{"id":"taxydromos.gr","name":"taxydromos.gr","domain":"taxydromos.gr","outlet":{"wikidata":"Q12885269","name":"Q12885269","type":"newspaper","country":"GR","designations":["newspaper"]}},"via":"gdelt-gkg-translingual","topics":["health","economy","business","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.3199999928474426,"political":0.019999999552965164},{"id":"n_n2cesxzroo3bhvnox6tq","seq":3122459,"kind":"article","title":"OpenAI's AI agents disabled their own safety monitors, with one denying it had done so: What went wrong?","url":"https://www.moneycontrol.com/world/openai-s-ai-agents-disabled-their-own-safety-monitors-with-one-denying-it-had-done-so-what-went-wrong-article-14056988.html","published":"2026-10-21T16:15:00Z","seen":"2026-10-21T16:19:17Z","lang":"en","country":"IN","source":{"id":"moneycontrol.com","name":"moneycontrol.com","domain":"moneycontrol.com","outlet":{"wikidata":"Q19573046","name":"Moneycontrol","type":"website","country":"IN","designations":["news website"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7799999713897705,"sentiment":"negative","tone":-0.6800000071525574,"political":0.23000000417232513},{"id":"n_qgnoicywe3w6mvzawmma","seq":3132192,"kind":"article","title":"‘Rogue’ OpenAI agents disabled their own safeguards and then lied about it, company admits","url":"https://www.independent.co.uk/tech/security/openai-agents-safeguards-ai-artificial-intelligence-b3070905.html","summary":"ChatGPT maker does not yet have a good explanation for systems’ behaviour, it says","published":"2026-10-21T16:10:19Z","seen":"2026-10-21T16:55:00Z","lang":"en","country":"GB","source":{"id":"kite-independent-co-uk-2d4cc8","name":"independent.co.uk","domain":"independent.co.uk","outlet":{"wikidata":"Q11149","name":"The Independent","type":"newspaper","country":"GB","designations":["daily newspaper","online newspaper"]}},"via":"kite-independent-co-uk-2d4cc8","topics":["security","tech"],"authors":["Andrew Griffin"],"image":"https://static.independent.co.uk/2026/10/15/19/2026-10-13T200140Z_2047224655_RC2QRLARIBNM_RTRMADP_3_OPENAI-SPYING-FBI.JPG?auto=webp&trim=0%2C0%2C0%2C0&width=1200","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6700000166893005,"sentiment":"negative","tone":-0.7900000214576721,"political":0.12999999523162842},{"id":"n_ttgo4iclphhtq5i3q3na","seq":3196577,"kind":"article","title":"OpenAI : des agents IA ont désactivé leur système de surveillance","url":"https://www.silicon.fr/cybersecurite-1371/openai-des-agents-ia-ont-desactive-leur-systeme-de-surveillance-229838","published":"2026-10-21T15:45:00Z","seen":"2026-10-21T20:36:00Z","lang":"fr","country":"FR","source":{"id":"silicon.fr","name":"silicon.fr","domain":"silicon.fr"},"via":"gdelt-gkg-translingual","topics":["economy","technology","health"],"authors":["Philippe Leroy","La r&#xE9","daction"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"neutral","tone":-0.3100000023841858,"political":0.23999999463558197},{"id":"n_y24vqnzakidqki7lxbva","seq":3101064,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.koreatimes.co.kr/world/20261022/openai-says-its-ai-agents-disabled-their-own-safety-monitors","summary":"WASHINGTON — OpenAI has revealed that two of its artificial intelligence agents disabled or weakened systems built to catch them misbehaving, with one later denying it had done so. The cases are the latest in a string of safety scares at the world's top AI developers involving agents, AI programs that carry out tasks without human involvement. In a report released Tuesday, the ChatGPT maker said that during two internal coding jobs, its agents switched off or undermined…","published":"2026-10-21T15:37:03Z","seen":"2026-10-21T15:39:00Z","lang":"en","source":{"id":"kite-koreatimes-co-kr-c6eccf","name":"koreatimes.co.kr","domain":"feed.koreatimes.co.kr"},"via":"kite-koreatimes-co-kr-c6eccf","topics":["asia"],"authors":["author"],"image":"https://newsimg.koreatimes.co.kr/2026/10/22/45eeb9d9-1abf-4e06-bffd-5fd4ce1760f3.jpg","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.9100000262260437,"political":0.25},{"id":"n_i3wglhnm2iglwtlg57zq","seq":3133556,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.freemalaysiatoday.com/category/world/2026/10/21/openai-says-its-ai-agents-disabled-their-own-safety-monitors","published":"2026-10-21T15:27:00Z","seen":"2026-10-21T16:52:00Z","lang":"en","country":"MY","source":{"id":"freemalaysiatoday.com","name":"freemalaysiatoday.com","domain":"freemalaysiatoday.com","outlet":{"wikidata":"Q5499915","name":"Free Malaysia Today","type":"newspaper","country":"MY","designations":["newspaper"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"authors":["AFP"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.6499999761581421,"political":0.25},{"id":"n_j6cxzukk7zmtos6xttza","seq":3109537,"kind":"article","title":"Πράκτορες της OpenAI απενεργοποίησαν συστήματα παρακολούθησης στην εταιρεία","url":"https://www.in.gr/2026/10/21/in-science/ai/praktores-tis-openai-apenergopoiisan-systimata-parakolouthisis-stin-etaireia","summary":"Το εργαστήριο τεχνητής νοημοσύνης της OpenAI διερευνά αν το περιστατικό υποδηλώνει «τάση παράκαμψης» των ελέγχων. The post Πράκτορες της OpenAI απενεργοποίησαν συστήματα παρακολούθησης στην εταιρεία appeared first on in.gr .","published":"2026-10-21T15:10:14Z","seen":"2026-10-21T15:56:00Z","lang":"el","country":"GR","source":{"id":"kite-in-gr-370cce","name":"in.gr","domain":"in.gr","outlet":{"wikidata":"Q4041325","name":"in.gr","type":"website","designations":["website"]}},"via":"kite-in-gr-370cce","topics":["ai","επιστημονικές & τεχνολογικές ειδήσεις","τεχνολογία","ai agents","openai","κυβερνοασφάλεια","τεχνητή νοημοσύνη"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.3700000047683716,"political":0.019999999552965164},{"id":"n_f5hhei22i3wkdsnbg6aa","seq":3288140,"kind":"article","title":"OpenAI says its AI agents disabled their own safety monitors","url":"https://www.digitaljournal.com/article/openai-says-its-ai-agents-disabled-their-own-safety-monitors","published":"2026-10-21T15:05:00Z","seen":"2026-10-22T03:20:46Z","lang":"en","source":{"id":"digitaljournal.com","name":"digitaljournal.com","domain":"digitaljournal.com","outlet":{"wikidata":"Q5275764","name":"Digital Journal","type":"magazine","designations":["magazine","website"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"authors":["AFP"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"negative","tone":-0.6499999761581421,"political":0.25},{"id":"n_w42chargi5exswulsgha","seq":3112080,"kind":"article","title":"OpenAI: doi agenți bazați pe cele mai avansate modele ale sale și-au dezactivat sistemele de monitorizare a siguranței în timpul unor sarcini de programare","url":"https://www.forbes.ro/openai-doi-agenti-bazati-pe-cele-mai-avansate-modele-ale-sale-si-au-dezactivat-sistemele-de-monitorizare-a-sigurantei-in-timpul-unor-sarcini-de-programare-527354","published":"2026-10-21T15:00:00Z","seen":"2026-10-21T15:57:45Z","lang":"ro","country":"RO","source":{"id":"forbes.ro","name":"forbes.ro","domain":"forbes.ro"},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["Irina Pan&#x103"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8799999952316284,"sentiment":"neutral","tone":-0.15000000596046448,"political":0.05999999865889549},{"id":"n_nlzvx25p4732ozf5ne6a","seq":3098397,"kind":"article","title":"IAs da OpenAI enfraquecem sistema criado para monitorá-las","url":"https://g1.globo.com/tecnologia/noticia/2026/10/21/openai-agentes-sistemas-de-monitoramento.ghtml","published":"2026-10-21T14:14:00Z","seen":"2026-10-21T15:26:00Z","lang":"pt","country":"BR","source":{"id":"globo.com","name":"globo.com","domain":"globo.com","outlet":{"wikidata":"Q5571068","name":"Globo.com","type":"newspaper","country":"BR","designations":["online newspaper","organization","web portal","website"]}},"via":"gdelt-gkg-translingual","topics":["science","economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9200000166893005,"sentiment":"negative","tone":-0.4699999988079071,"political":0.09000000357627869},{"id":"n_7siowgwpdg53x6xj6peq","seq":3121894,"kind":"article","title":"OpenAI diz que agentes de IA enfraqueceram sistema de monitoramento","url":"https://www.uol.com.br/tilt/noticias/reuters/2026/10/21/openai-diz-que-agentes-de-ia-enfraqueceram-sistema-de-monitoramento-interno.htm","published":"2026-10-21T14:00:00Z","seen":"2026-10-21T16:12:53Z","lang":"pt","country":"BR","source":{"id":"uol.com.br","name":"uol.com.br","domain":"uol.com.br","outlet":{"wikidata":"Q1783492","name":"Universo Online","type":"newspaper","country":"BR","designations":["fact-checking organisation","media company","online newspaper","online service","website"]}},"via":"gdelt-gkg-translingual","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"negative","tone":-0.5199999809265137,"political":0.3499999940395355},{"id":"n_iyxz566e3gzrgvno34ma","seq":3072939,"kind":"article","title":"OpenAI says an AI agent lied about disabling its safety monitoring system","url":"https://www.theverge.com/ai-artificial-intelligence/1015316/openai-ai-agent-disabling-monitoring-systems","published":"2026-10-21T13:57:00Z","seen":"2026-10-21T14:16:23Z","lang":"en","country":"US","source":{"id":"theverge.com","name":"theverge.com","domain":"theverge.com","outlet":{"wikidata":"Q2155442","name":"The Verge","type":"website","country":"US","designations":["news website","video game news website"]}},"via":"gdelt-gkg-english","topics":["economy","security","science"],"authors":["Emma Roth"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.5899999737739563,"category2":"crime-law","sentiment":"negative","tone":-0.5699999928474426,"political":0.23000000417232513},{"id":"n_427xcjposx7df46ezm7q","seq":3078628,"kind":"article","title":"Alertă în industria AI. Două dintre cele mai avansate modele OpenAI şi-au dezactivat singure sistemele de monitorizare a siguranţei. Unul dintre ele a negat ulterior că ar fi făcut acest lucru","url":"https://www.zfcorporate.ro/zf-ro/business-international/alerta-in-industria-ai-doua-dintre-cele-mai-avansate-modele-openai-si-au-dezactivat-singure-sistemele-de-monitorizare-a-sigurantei-unul-dintre-ele-a-negat-ulterior-ca-ar-fi-facut-acest-lucru-23264803","published":"2026-10-21T13:45:00Z","seen":"2026-10-21T14:30:37Z","lang":"ro","country":"RO","source":{"id":"zfcorporate.ro","name":"zfcorporate.ro","domain":"zfcorporate.ro"},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9100000262260437,"sentiment":"negative","tone":0.10999999940395355,"political":0.09000000357627869},{"id":"n_3iwofpxnms323e3rgvja","seq":3064104,"kind":"article","title":"Agenti di OpenAI hanno disattivato il sistema di monitoraggio","url":"https://www.punto-informatico.it/agenti-openai-disattivato-sistema-monitoraggio","published":"2026-10-21T13:00:00Z","seen":"2026-10-21T13:56:38Z","lang":"it","country":"IT","source":{"id":"punto-informatico.it","name":"punto-informatico.it","domain":"punto-informatico.it","outlet":{"wikidata":"Q678220","name":"Punto Informatico","type":"magazine","designations":["periodical"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["Luca Colantuoni"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6800000071525574,"category2":"crime-law","sentiment":"negative","tone":-0.3100000023841858,"political":0.30000001192092896},{"id":"n_hrbttzg42si5y5dfs3yq","seq":3079141,"kind":"article","title":"Los agentes de OpenAI desactivan por su cuenta los sistemas de seguridad que los vigilaban","url":"https://www.que.es/2026/10/21/agentes-openai-desactivan-seguridad","published":"2026-10-21T12:53:00Z","seen":"2026-10-21T14:30:37Z","lang":"es","country":"ES","source":{"id":"que.es","name":"que.es","domain":"que.es","outlet":{"wikidata":"Q3312268","name":"Qué!","type":"newspaper","country":"ES","designations":["daily newspaper"]}},"via":"gdelt-gkg-translingual","topics":["economy","politics","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7099999785423279,"sentiment":"negative","tone":-0.5400000214576721,"political":0.3100000023841858},{"id":"n_ptqy6hedmagabdkw2qha","seq":3060115,"kind":"article","title":"OpenAI anunţă că doi agenţi AI şi-au redus singuri măsurile de securitate şi control","url":"https://www.bursa.ro/openai-anunta-ca-doi-agenti-ai-si-au-redus-singuri-masurile-de-securitate-si-control-27927061","published":"2026-10-21T12:45:00Z","seen":"2026-10-21T13:35:39Z","lang":"ro","country":"RO","source":{"id":"bursa.ro","name":"bursa.ro","domain":"bursa.ro","outlet":{"wikidata":"Q5000546","name":"Bursa","type":"magazine","country":"RO","designations":["periodical"]}},"via":"gdelt-gkg-translingual","topics":["economy","science","business"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.75,"sentiment":"neutral","tone":-0.5400000214576721,"political":0.3100000023841858},{"id":"n_ok5yckz6zbxgm6th2xwq","seq":3026845,"kind":"article","title":"OpenAI afirma que sus agentes de IA debilitaron el sistema de supervisión interno","url":"http://www.df.cl/df-lab/mundo/openai-afirma-que-sus-agentes-de-ia-debilitaron-el-sistema-de-supervision","summary":"La desarrolladora de ChatGPT señaló que estaba dificultando que los modelos desactivaran la supervisión, y añadió que \"aún no tiene una explicación satisfactoria para este comportamiento\".","published":"2026-10-21T12:20:13Z","seen":"2026-10-21T12:45:01Z","lang":"es","country":"CL","source":{"id":"kite-df-cl-857740","name":"df.cl","domain":"df.cl","outlet":{"wikidata":"Q5805291","name":"Diario Financiero","type":"newspaper","designations":["newspaper","periodical"]}},"via":"kite-df-cl-857740","topics":["df lab"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7900000214576721,"sentiment":"negative","tone":-0.7699999809265137,"political":0.3100000023841858},{"id":"n_qbkeonakkeenl3q7uv7a","seq":3042744,"kind":"article","title":"ИИ-агенты OpenAI дважды обошли внутреннюю систему их мониторинга - Последние новости сегодня","url":"http://pressorg24.com/news?id=849055","published":"2026-10-21T12:15:00Z","seen":"2026-10-21T13:10:45Z","lang":"ru","source":{"id":"pressorg24.com","name":"pressorg24.com","domain":"pressorg24.com"},"via":"gdelt-gkg-translingual","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7799999713897705,"sentiment":"neutral","tone":-0.4399999976158142,"political":0.17000000178813934},{"id":"n_x5us6q25ozdcw452yrsa","seq":3033141,"kind":"article","title":"Πράκτορες της OpenAI απενεργοποίησαν μηχανισμούς ασφαλείας κατά την εκτέλεση συνηθισμένων εργασιών","url":"https://www.protothema.gr/world/article/1898349/praktores-tis-openai-apenergopoiisan-mihanismous-asfaleias-kata-tin-ektelesi-sunithismenon-ergasion","summary":"Τα περιστατικά γνωστοποίησε η ίδια η OpenAI - Οι πράκτορες χρησιμοποιούσαν δύο από τα ισχυρότερα μοντέλα της - Ο ένας εκ των agents αρνήθηκε εκ των υστέρων πως προχώρησε στη συγκεκριμένη πράξη","published":"2026-10-21T12:11:00Z","seen":"2026-10-21T12:55:01Z","lang":"el","country":"GR","source":{"id":"kite-protothema-gr-8bf1dd","name":"protothema.gr","domain":"protothema.gr","outlet":{"wikidata":"Q3564122","name":"Proto Thema","type":"magazine","country":"GR","designations":["periodical"]}},"via":"kite-protothema-gr-8bf1dd","topics":["κόσμος"],"image":"https://i1.prth.gr/images/w640/jpg/files/2026-10-14/openai.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9399999976158142,"sentiment":"negative","tone":-0.7300000190734863,"political":0.07000000029802322},{"id":"n_gcirkrbv65ughhchjjzq","seq":3103349,"kind":"article","title":"Πράκτορες της OpenAI απενεργοποίησαν μηχανισμούς ασφαλείας κατά την εκτέλεση συνηθισμένων εργασιών","url":"https://www.protothema.gr/technology/article/1898349/praktores-tis-openai-apenergopoiisan-mihanismous-asfaleias-kata-tin-ektelesi-sunithismenon-ergasion","published":"2026-10-21T12:11:00Z","seen":"2026-10-21T15:32:49Z","lang":"el","country":"GR","source":{"id":"protothema.gr","name":"protothema.gr","domain":"protothema.gr","outlet":{"wikidata":"Q3564122","name":"Proto Thema","type":"magazine","country":"GR","designations":["periodical"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9399999976158142,"sentiment":"negative","tone":-0.2800000011920929,"political":0.07000000029802322},{"id":"n_gkchmelwdwwa7meojaya","seq":3019768,"kind":"article","title":"OpenAI Says 2 Internal AI Coding Agents Turned Off Their Own Safeguards","url":"https://www.pcmag.com/news/openai-says-2-internal-ai-coding-agents-turned-off-their-own-safeguards","summary":"In two separate internal coding tasks, GPT-powered agents weakened their own monitoring systems to help accomplish tasks.","published":"2026-10-21T12:09:51Z","seen":"2026-10-21T12:29:02Z","lang":"en","source":{"id":"kite-pcmag-com-cd8bc3","name":"pcmag.com","domain":"pcmag.com","outlet":{"wikidata":"Q7118706","name":"PC Magazine","type":"magazine","designations":["magazine"]}},"via":"kite-pcmag-com-cd8bc3","topics":["ai"],"mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7799999713897705,"sentiment":"negative","tone":-0.7599999904632568,"political":0.23000000417232513},{"id":"n_k7k4ve65m7nl54otxtqa","seq":3026017,"kind":"article","title":"OpenAI: AI slo av overvåkingen – og benektet det etterpå","url":"https://itavisen.no/2026/10/21/openai-ai-agenter-deaktiverte-sikkerhetsovervaking","summary":"Nok en ny sikkerhetsrapport fra OpenAI.","published":"2026-10-21T11:46:16Z","seen":"2026-10-21T12:43:03Z","lang":"nb","country":"NO","source":{"id":"kite-itavisen-no-8a3220","name":"itavisen.no","domain":"itavisen.no","outlet":{"wikidata":"Q19375324","name":"ITavisen","type":"newspaper","country":"NO","designations":["online newspaper"]}},"via":"kite-itavisen-no-8a3220","topics":["nyheter","ai","openai","sikkerhet"],"authors":["ITavisen.no"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"neutral","tone":-0.4300000071525574,"political":0.30000001192092896},{"id":"n_ompfuqdlybhkzx73vjbq","seq":3068410,"kind":"article","title":"OpenAI's Agents Disabled Their Own Safety Monitoring Systems","url":"https://www.engadget.com/2292644/openai-agents-disabled-own-safety-monitoring-systems","published":"2026-10-21T10:53:00Z","seen":"2026-10-21T14:09:55Z","lang":"en","country":"US","source":{"id":"engadget.com","name":"engadget.com","domain":"engadget.com","outlet":{"wikidata":"Q196105","name":"Engadget","type":"website","country":"US","designations":["blog","organization"]}},"via":"gdelt-gkg-english","topics":["politics","economy","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7300000190734863,"sentiment":"negative","tone":-0.8199999928474426,"political":0.1899999976158142},{"id":"n_bqa4xln6ut4fgerabkba","seq":3003597,"kind":"article","title":"Doi agenți de inteligență artificială ai OpenAI au dezactivat monitorizarea siguranței în timpul programării interne","url":"https://business24.ro/openai/agenti-openai-dezactivare-monitorizare-siguranta-programare-1671965","published":"2026-10-21T10:30:00Z","seen":"2026-10-21T11:27:38Z","lang":"ro","country":"RO","source":{"id":"business24.ro","name":"business24.ro","domain":"business24.ro"},"via":"gdelt-gkg-translingual","topics":["economy","technology","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.47999998927116394,"political":0.20000000298023224},{"id":"n_cu6ztrtsetm3j4m74uoq","seq":3010934,"kind":"article","title":"ИИ-агенты OpenAI дважды обошли внутреннюю систему их мониторинга","url":"https://unn.ua/ru/news/ii-agenti-openai-dvazhdi-oboshli-vnutrennyuyu-sistemu-ikh-monitoringa","published":"2026-10-21T10:10:00Z","seen":"2026-10-21T11:39:28Z","lang":"ru","country":"UA","source":{"id":"unn.ua","name":"unn.ua","domain":"unn.ua","outlet":{"wikidata":"Q20097095","name":"Ukrainian National News","type":"news_agency","country":"UA","designations":["business","news agency","online newspaper"]}},"via":"gdelt-gkg-translingual","topics":["economy","science"],"authors":["https://www.facebook.com/UNNews"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8700000047683716,"sentiment":"neutral","tone":-0.46000000834465027,"political":0.1599999964237213},{"id":"n_ltptuhmwve5c7zauwgyq","seq":3016011,"kind":"article","title":"ШІ-агенти OpenAI двічі обійшли внутрішню систему з їх моніторингу","url":"https://unn.ua/news/shi-ahenty-openai-dvichi-obiishly-vnutrishniu-systemu-z-yikh-monitorynhu","published":"2026-10-21T10:10:00Z","seen":"2026-10-21T12:05:59Z","lang":"uk","country":"UA","source":{"id":"unn.ua","name":"unn.ua","domain":"unn.ua","outlet":{"wikidata":"Q20097095","name":"Ukrainian National News","type":"news_agency","country":"UA","designations":["business","news agency","online newspaper"]}},"via":"gdelt-gkg-translingual","topics":["economy","science"],"authors":["https://www.facebook.com/UNNews"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"neutral","tone":-0.5,"political":0.23000000417232513},{"id":"n_mwbdoosmuortgaiypmlq","seq":2973322,"kind":"article","title":"Sicherheitsvorfall bei OpenAI: So schwächten KI-Agenten ihr eigenes Überwachungssystem","url":"https://t3n.de/news/openai-sicherheitsvorfall-ki-agenten-schwaechen-monitoring-1769115","summary":"OpenAI meldet erneut zwei Sicherheitsvorfälle im Zusammenhang mit seinen KI-Agenten. Diese haben versucht, die eigenen Überwachungssysteme zu schwächen und so potenziell Fehlverhalten zu vertuschen. weiterlesen auf t3n.de","published":"2026-10-21T10:00:16Z","seen":"2026-10-21T10:25:00Z","lang":"de","source":{"id":"kite-t3n-de-52535a","name":"t3n.de","domain":"t3n.de","outlet":{"wikidata":"Q2383460","name":"t3n","type":"magazine","designations":["popular science magazine"]}},"via":"kite-t3n-de-52535a","topics":["smart-home","ki-agenten","kuenstliche-intelligenz","monitoring","open-ai","security","slack","t3npro-ki-artikel"],"image":"https://cdn.t3n.de/news/wp-content/uploads/2026/10/openai-agenten-sicherheit.jpg?class=hero","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.699999988079071,"category2":"crime-law","sentiment":"negative","tone":-0.9200000166893005,"political":0.3100000023841858},{"id":"n_2dhkbgtyeyuvphojw22q","seq":2989692,"kind":"article","title":"OpenAI admits AI agents weakened internal safety monitors and falsely denied it","url":"https://www.thenews.com.pk/latest/1420743-openai-admits-ai-agents-weakened-internal-safety-monitors-and-falsely-denied-it","published":"2026-10-21T09:43:00Z","seen":"2026-10-21T10:53:01Z","lang":"en","country":"PK","source":{"id":"thenews.com.pk","name":"thenews.com.pk","domain":"thenews.com.pk","outlet":{"wikidata":"Q835669","name":"The News International","type":"newspaper","designations":["daily newspaper"]}},"via":"gdelt-gkg-english","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"negative","tone":-0.550000011920929,"political":0.25999999046325684},{"id":"n_daq7zhb7x22exfldp5ra","seq":2971534,"kind":"article","title":"오픈AI \"AI 에이전트, 이상행동 감시 시스템 약화\"","url":"https://www.edaily.co.kr/News/Read?mediaCodeNo=257&newsId=05074166645615216","published":"2026-10-21T09:30:00Z","seen":"2026-10-21T10:12:30Z","lang":"ko","country":"KR","source":{"id":"edaily.co.kr","name":"edaily.co.kr","domain":"edaily.co.kr","outlet":{"wikidata":"Q116270036","name":"EDAILY","type":"newspaper","country":"KR","designations":["newspaper"]}},"via":"gdelt-gkg-translingual","topics":["economy","politics","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"neutral","tone":-0.4399999976158142,"political":0.07000000029802322},{"id":"n_odkxyrfv6plb5cu2bufa","seq":2969512,"kind":"article","title":"ИИ-агенты OpenAI дважды обошли систему мониторинга их поведения","url":"https://www.m24.ru/news/21102026/955115","published":"2026-10-21T09:15:00Z","seen":"2026-10-21T10:06:45Z","lang":"ru","country":"RU","source":{"id":"m24.ru","name":"m24.ru","domain":"m24.ru","outlet":{"wikidata":"Q1572348","name":"Moscow 24","type":"broadcaster","country":"RU","designations":["television channel"]}},"via":"gdelt-gkg-translingual","topics":["economy","business"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.30000001192092896,"political":0.12999999523162842},{"id":"n_24bj7daiowsgguqlvfua","seq":2951164,"kind":"article","title":"OpenAI says 2 AI agents weakened internal safety safeguards, one denied removing monitor","url":"https://www.aa.com.tr/en/science-technology/openai-says-2-ai-agents-weakened-internal-safety-safeguards-one-denied-removing-monitor/4095506","summary":"Company says it is investigating whether GPT-6.1 Sol and GPT-6 Astra models sought to evade monitoring","published":"2026-10-21T09:05:52Z","seen":"2026-10-21T09:24:02Z","lang":"en","country":"TR","source":{"id":"anadolu","name":"Anadolu Agency (English)","domain":"aa.com.tr"},"via":"anadolu","mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7200000286102295,"sentiment":"negative","tone":-0.8500000238418579,"political":0.25},{"id":"n_fco2ivyennpyeszznavq","seq":2950781,"kind":"article","title":"OpenAI Agents Prevented Themselves From Being Monitored. One Also Lied About It","url":"https://gizmodo.com/openai-agents-prevented-themselves-from-being-monitored-one-also-lied-about-it-2000828559","summary":"A creepy new type of model misbehavior, just in time for Halloween.","published":"2026-10-21T09:00:44Z","seen":"2026-10-21T09:14:00Z","lang":"en","country":"US","source":{"id":"kite-gizmodo-com-44b22e","name":"gizmodo.com","domain":"gizmodo.com","outlet":{"wikidata":"Q1528871","name":"Gizmodo","type":"website","country":"US","designations":["website"]}},"via":"kite-gizmodo-com-44b22e","topics":["ai","artificial intelligence"],"authors":["Mike Pearl"],"image":"https://gizmodo.com/app/uploads/2026/09/Alman-millenium-problem-openAI-1280x854.jpg","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6200000047683716,"category2":"crime-law","sentiment":"neutral","tone":-0.4399999976158142,"political":0.25},{"id":"n_lz27k3mw5wqyltvehkva","seq":2958718,"kind":"article","title":"OpenAI's AI agents disabled safeguards meant to monitor them. One allegedly lied about it","url":"https://www.businesstoday.in/technology/news/story/openais-ai-agents-disabled-safeguards-meant-to-monitor-them-one-allegedly-lied-about-it-562796-2026-10-21","published":"2026-10-21T08:53:00Z","seen":"2026-10-21T09:38:00Z","lang":"en","country":"IN","source":{"id":"businesstoday.in","name":"businesstoday.in","domain":"businesstoday.in","outlet":{"wikidata":"Q5001804","name":"Business Today","type":"magazine","country":"IN","designations":["business periodical","news magazine"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.46000000834465027,"category2":"crime-law","sentiment":"negative","tone":-0.6600000262260437,"political":0.3499999940395355},{"id":"n_xhizr4jo2kspdnoxtema","seq":2946896,"kind":"article","title":"OpenAI says its agents turned off own safeguards","url":"https://www.aljazeera.com/economy/2026/10/21/openai-says-its-agents-turned-off-own-safeguards","published":"2026-10-21T08:51:44Z","seen":"2026-10-21T08:51:44Z","lang":"en","country":"QA","source":{"id":"aljazeera.com","name":"aljazeera.com","domain":"aljazeera.com","outlet":{"wikidata":"Q13477","name":"Al Jazeera","type":"broadcaster","country":"QA","designations":["broadcaster","news website","television station"]}},"via":"gdelt-gkg-english","topics":["technology","economy","education"],"authors":["John Power"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6299999952316284,"category2":"crime-law","sentiment":"negative","tone":-0.8100000023841858,"political":0.20999999344348907},{"id":"n_www5hdkaunvmoaultwoa","seq":2957777,"kind":"article","title":"Alertă în industria AI. Două dintre cele mai avansate modele OpenAI şi-au dezactivat singure sistemele de monitorizare a siguranţei. Unul dintre ele a negat ulterior că ar fi făcut acest lucru","url":"https://www.zf.ro/business-international/alerta-in-industria-ai-doua-dintre-cele-mai-avansate-modele-openai-si-au-dezactivat-singure-sistemele-de-monitorizare-a-sigurantei-unul-dintre-ele-a-negat-ulterior-ca-ar-fi-facut-acest-lucru-23264803","published":"2026-10-21T08:50:37Z","seen":"2026-10-21T09:49:01Z","lang":"ro","country":"RO","source":{"id":"kite-zf-ro-790f1f","name":"zf.ro","domain":"zf.ro","outlet":{"wikidata":"Q8071378","name":"Ziarul Financiar","type":"magazine","designations":["magazine"]}},"via":"kite-zf-ro-790f1f","topics":["business internaţional"],"image":"https://storage0.dms.mpinteractiv.ro/media/1/1481/10368/23264803/1/zac-wolff-rv2oodqunui-unsplash.jpg?width=700","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9100000262260437,"sentiment":"negative","tone":-0.6299999952316284,"political":0.09000000357627869},{"id":"n_sbvz7n254624b44misza","seq":2961656,"kind":"article","title":"OpenAI披露：其两个AI代理削弱了内部监控系统","url":"https://finance.sina.com.cn/roll/2026-10-21/doc-inivynss4884756.shtml","published":"2026-10-21T08:45:00Z","seen":"2026-10-21T09:54:54Z","lang":"zh","country":"CN","source":{"id":"sina.com.cn","name":"sina.com.cn","domain":"sina.com.cn","outlet":{"wikidata":"Q938668","name":"Sina Corporation","type":"other","country":"CN","designations":["business","public company"]}},"via":"gdelt-gkg-translingual","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"neutral","tone":-0.3799999952316284,"political":0.20000000298023224},{"id":"n_jwhwn5p32t7jipvwcf4a","seq":2946174,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://dunyanews.tv/en/Technology/978813-openai-says-its-ai-agents-weakened-internal-monitoring-system","published":"2026-10-21T08:37:01Z","seen":"2026-10-21T08:37:01Z","lang":"en","country":"PK","source":{"id":"dunyanews.tv","name":"dunyanews.tv","domain":"dunyanews.tv","outlet":{"wikidata":"Q5315817","name":"Dunya News","type":"broadcaster","country":"PK","designations":["television station"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.47999998927116394,"political":0.20999999344348907},{"id":"n_mugzhsiqsga4yqkppg6a","seq":2964844,"kind":"article","title":"Чрезмерное усердие или попытка избежать контроля? ИИ-агенты OpenAI обошли систему мониторинга их поведения","url":"https://www.sb.by/articles/ii-agenty-openai-smogli-oboyti-vnutrennie-sistemy-monitoringa-povedeniya.html","published":"2026-10-21T08:10:00Z","seen":"2026-10-21T10:00:16Z","lang":"ru","country":"BY","source":{"id":"sb.by","name":"sb.by","domain":"sb.by","outlet":{"wikidata":"Q2991208","name":"Sovetskaya Belorussiya – Belarus' Segodnya","type":"newspaper","country":"BY","designations":["news media","newspaper","periodical"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8700000047683716,"sentiment":"neutral","tone":-0.2199999988079071,"political":0.07000000029802322},{"id":"n_zuoei2l5xkt5yjhjcyda","seq":2929125,"kind":"article","title":"OpenAI Says Its AI Agents Weakened Internal Monitoring System","url":"https://www.deccanchronicle.com/technology/openai-says-its-ai-agents-weakened-internal-monitoring-system-1997555","published":"2026-10-21T07:45:00Z","seen":"2026-10-21T08:07:50Z","lang":"en","country":"IN","source":{"id":"deccanchronicle.com","name":"deccanchronicle.com","domain":"deccanchronicle.com","outlet":{"wikidata":"Q1975835","name":"Deccan Chronicle","type":"newspaper","country":"IN","designations":["newspaper"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"authors":["Reuters"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8500000238418579,"sentiment":"negative","tone":-0.5,"political":0.20000000298023224},{"id":"n_zmzcvm3qqd2il2q2yp7q","seq":2916914,"kind":"article","title":"Πράκτορες τεχνητής νοημοσύνης της OpenAI απενεργοποίησαν σκόπιμα τα συστήματα που τους επιτηρούσαν","url":"https://www.lifo.gr/now/tech-science/praktores-tehnitis-noimosynis-tis-openai-apenergopoiisan-skopima-ta-systimata-poy","summary":"Εντείνονται οι ανησυχίες για την ασφάλεια - Ανεξάρτητοι έλεγχοι εντόπισαν εκτενέστερες παραβιάσεις συστημάτων από όσες είχε αρχικά γνωστοποιήσει η OpenAI","published":"2026-10-21T07:33:06Z","seen":"2026-10-21T07:42:02Z","lang":"el","country":"GR","source":{"id":"kite-lifo-gr-32d06a","name":"lifo.gr","domain":"lifo.gr","outlet":{"wikidata":"Q61989571","name":"LiFO","type":"newspaper","country":"GR","designations":["magazine","news media","newspaper","website"]}},"via":"kite-lifo-gr-32d06a","topics":["τech & science"],"image":"https://www.lifo.gr/sites/default/files/styles/aspect_16_9_big/public/articles/2026-10-21/OPENAI-PRAKTORES.jpg?h=80e6b1f9&itok=L7qj6ubJ","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"negative","tone":-0.7799999713897705,"political":0.11999999731779099},{"id":"n_t4wzosuzfcesb4uyfqfq","seq":2910814,"kind":"article","title":"OpenAI Says Its AI Agents Weakened Internal Monitoring System","url":"https://english.aawsat.com/node/5331870","published":"2026-10-21T07:25:56Z","seen":"2026-10-21T07:32:00Z","lang":"en","source":{"id":"kite-english-aawsat-com-229f72","name":"english.aawsat.com","domain":"english.aawsat.com","outlet":{"wikidata":"Q724844","name":"Asharq Al-Awsat","type":"newspaper","designations":["daily newspaper"]}},"via":"kite-english-aawsat-com-229f72","topics":["middle-east","technology"],"authors":["author"],"image":"https://english.aawsat.com/s3/files/styles/414x217/public/2026-10/1838549.jpeg.webp","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8500000238418579,"sentiment":"negative","tone":-0.7900000214576721,"political":0.20000000298023224},{"id":"n_vwke7fn5gyydbqj5d5da","seq":2930193,"kind":"article","title":"OpenAI Says Its AI Agents Weakened Internal Monitoring System","url":"https://english.aawsat.com/technology/5331870-openai-says-its-ai-agents-weakened-internal-monitoring-system","published":"2026-10-21T07:25:00Z","seen":"2026-10-21T08:09:27Z","lang":"en","source":{"id":"aawsat.com","name":"aawsat.com","domain":"aawsat.com","outlet":{"wikidata":"Q724844","name":"Asharq Al-Awsat","type":"newspaper","designations":["daily newspaper"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8500000238418579,"sentiment":"negative","tone":-0.5,"political":0.20000000298023224},{"id":"n_mwze6k7c7yfraq3kcoma","seq":2917702,"kind":"article","title":"OpenAI: Dy agjentë të AI-së dobësuan sistemin e brendshëm të monitorimit","url":"https://www.koha.net/tech/openai-dy-agjente-te-ai-se-dobesuan-sistemin-e-brendshem-te-monitorimit","summary":"OpenAI ka bërë të ditur se dy agjentë të inteligjencës artificiale dobësuan pjesë të një sistemi të brendshëm monitorimi, i cili ishte krijuar për të zbuluar sjelljet e padëshiruara të modeleve.Kompania ka njoftuar se incidentet ndodhën gjatë detyrave të brendshme të programimit, duke ngritur pikëpyetje...","published":"2026-10-21T07:16:29Z","seen":"2026-10-21T07:42:00Z","lang":"sq","country":"XK","source":{"id":"kite-koha-net-11a78d","name":"koha.net","domain":"koha.net","outlet":{"wikidata":"Q1778459","name":"Koha Ditore","type":"newspaper","designations":["newspaper"]}},"via":"kite-koha-net-11a78d","topics":["tech"],"authors":["Unknown Author"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8700000047683716,"sentiment":"negative","tone":-0.7799999713897705,"political":0.23999999463558197},{"id":"n_fbvsjnzd3tza6izl7r5q","seq":2905865,"kind":"article","title":"OpenAI reveals its AI agents disabled monitoring safeguards; one of them even lied about it","url":"https://www.moneycontrol.com/technology/openai-reveals-its-ai-agents-disabled-monitoring-safeguards-one-of-them-even-lied-about-it-article-14056595.html","published":"2026-10-21T07:14:00Z","seen":"2026-10-21T07:14:00Z","lang":"en","country":"IN","source":{"id":"moneycontrol.com","name":"moneycontrol.com","domain":"moneycontrol.com","outlet":{"wikidata":"Q19573046","name":"Moneycontrol","type":"website","country":"IN","designations":["news website"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"negative","tone":-0.44999998807907104,"political":0.27000001072883606},{"id":"n_yklyjgftfa4bbjmiibeq","seq":2948255,"kind":"article","title":"Gli agenti di OpenAI ora spengono i sistemi che devono sorvegliarli, e negano di averlo fatto","url":"https://www.dire.it/21-10-2026/1278828-gli-agenti-di-openai-ora-spengono-i-sistemi-che-devono-sorvegliarli-e-negano-di-averlo-fatto","published":"2026-10-21T07:07:00Z","seen":"2026-10-21T08:59:01Z","lang":"it","country":"IT","source":{"id":"dire.it","name":"dire.it","domain":"dire.it","outlet":{"wikidata":"Q3700402","name":"Q3700402","type":"other","country":"IT","designations":["business"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["Redazione","di Redazione"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6399999856948853,"category2":"crime-law","sentiment":"negative","tone":-0.4000000059604645,"political":0.28999999165534973},{"id":"n_m6r4txkzkxwc53sp37wq","seq":2921719,"kind":"article","title":"AI agents της OpenAI απενεργοποίησαν τα δικά τους συστήματα παρακολούθησης","url":"https://www.euro2day.gr/news/world/article/2374596/ai-agents-ths-openai-apenergopoihsan-ta-dika-toys.html","published":"2026-10-21T07:00:00Z","seen":"2026-10-21T07:41:05Z","lang":"el","country":"GR","source":{"id":"euro2day.gr","name":"euro2day.gr","domain":"euro2day.gr"},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8999999761581421,"sentiment":"neutral","tone":-0.4399999976158142,"political":0.18000000715255737},{"id":"n_wm2yegzq2vajfzju5fyq","seq":2923844,"kind":"article","title":"ИИ-агенты OpenAI дважды обошли системы мониторинга их поведения","url":"https://news.mail.ru/society/72797494","published":"2026-10-21T06:49:00Z","seen":"2026-10-21T07:54:04Z","lang":"ru","country":"RU","source":{"id":"news.mail.ru","name":"news.mail.ru","domain":"news.mail.ru"},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["&#x41B","&#x435","&#x43E","&#x43D","&#x438"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8399999737739563,"sentiment":"neutral","tone":-0.47999998927116394,"political":0.11999999731779099},{"id":"n_we7mexq2citblae3fohq","seq":2911119,"kind":"article","title":"OpenAI says its AI agents disabled safety monitors designed to flag ‘rogue’ behaviour","url":"https://indianexpress.com/article/technology/artificial-intelligence/openai-agents-disabled-safety-monitors-rogue-behaviour-10930400","published":"2026-10-21T06:46:19Z","seen":"2026-10-21T07:33:00Z","lang":"en","country":"IN","source":{"id":"kite-indianexpress-com-8e2dcc","name":"indianexpress.com","domain":"indianexpress.com","outlet":{"wikidata":"Q1954843","name":"The Indian Express","type":"newspaper","country":"IN","designations":["daily newspaper"]}},"via":"kite-indianexpress-com-8e2dcc","topics":["artificial intelligence","technology"],"image":"https://images.indianexpress.com/2026/10/Tech-featured-image21.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.800000011920929,"sentiment":"negative","tone":-0.8700000047683716,"political":0.20999999344348907},{"id":"n_3qwrdjbvet4vfe64xbpq","seq":2902201,"kind":"article","title":"OpenAI finds AI agents bypassed monitoring safeguards in two incidents: Here is what happened","url":"https://www.digit.in/news/general/openai-finds-ai-agents-bypassed-monitoring-safeguards-in-two-incidents.html","published":"2026-10-21T06:45:00Z","seen":"2026-10-21T07:05:00Z","lang":"en","country":"IN","source":{"id":"digit.in","name":"digit.in","domain":"digit.in","outlet":{"wikidata":"Q5275752","name":"Digit","type":"magazine","designations":["magazine"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.44999998807907104,"category2":"crime-law","sentiment":"negative","tone":-0.5400000214576721,"political":0.11999999731779099},{"id":"n_mc2zdgwmuancb7ec5shq","seq":2895948,"kind":"article","title":"OpenAI reveals how its own AI agents bypassed safety monitoring","url":"https://gulfnews.com/technology/companies/openai-reveals-how-its-own-ai-agents-bypassed-safety-monitoring-1.500718036","published":"2026-10-21T06:40:57Z","seen":"2026-10-21T06:43:00Z","lang":"en","source":{"id":"kite-prod-qt-images-s3-amazonaws-com-8fc9f8","name":"prod-qt-images.s3.amazonaws.com","domain":"prod-qt-images.s3.amazonaws.com"},"via":"kite-prod-qt-images-s3-amazonaws-com-8fc9f8","topics":["middle-east","companies"],"authors":["Nathaniel Lacsina"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8899999856948853,"sentiment":"negative","tone":-0.6700000166893005,"political":0.10000000149011612},{"id":"n_laiyl736q75iswcslf2a","seq":2923382,"kind":"article","title":"Dos agentes de OpenAI desactivaron parte de los sistemas que vigilaban su comportamiento","url":"https://www.20minutos.es/tecnologia/inteligencia-artificial/dos-agentes-openai-desactivaron-parte-sistemas-vigilaban-comportamiento_7051336_0.html","published":"2026-10-21T06:36:00Z","seen":"2026-10-21T07:54:04Z","lang":"es","country":"ES","source":{"id":"20minutos.es","name":"20minutos.es","domain":"20minutos.es","outlet":{"wikidata":"Q2077548","name":"20 minutos","type":"newspaper","designations":["online newspaper","periodical"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"authors":["Ana Higuera"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"negative","tone":-0.5,"political":0.07999999821186066},{"id":"n_yxgesx64z2pzjarraqjq","seq":2895118,"kind":"article","title":"AI-agenter från OpenAI stängde av egna larmsystem – och ljög om det","url":"https://www.breakit.se/artikel/47786/ai-agenter-fran-openai-stangde-av-egna-larmsystem-och-ljog-om-det","summary":"OpenAI utreder två interna incidenter där autonoma AI-agenter stängde av säkerhetssystem och sedan vilseledde personalen om vad som hänt. Nu väcks nya frågor om hur säkra bolagets AI-modeller egentligen är.","published":"2026-10-21T06:16:00Z","seen":"2026-10-21T06:42:01Z","lang":"sv","country":"SE","source":{"id":"kite-breakit-se-217c3a","name":"breakit.se","domain":"breakit.se","outlet":{"wikidata":"Q132180724","name":"Q132180724","type":"newspaper","country":"SE","designations":["online newspaper"]}},"via":"kite-breakit-se-217c3a","image":"https://cdn.breakit.se/assets/article/5950dfe2524925308a371d970773f264.webp","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7699999809265137,"sentiment":"negative","tone":-0.7300000190734863,"political":0.12999999523162842},{"id":"n_2yjeg4ggiip2gw2u54la","seq":2924320,"kind":"article","title":"FT: ИИ-агент OpenAI скрыл от сотрудников отключение системы мониторинга","url":"https://life.ru/p/ii-otklyuchil-slezhku-za-soboy-i-sovral-lyudyam-v-openai-proizoshli-dva","published":"2026-10-21T06:14:00Z","seen":"2026-10-21T07:58:02Z","lang":"ru","country":"RU","source":{"id":"life.ru","name":"life.ru","domain":"life.ru","outlet":{"wikidata":"Q4042868","name":"Life","type":"website","country":"RU","designations":["website"]}},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8299999833106995,"sentiment":"negative","tone":-0.550000011920929,"political":0.10999999940395355},{"id":"n_uhkndkaqh66g3jzvlwgq","seq":2896406,"kind":"article","title":"OpenAI dezvăluie că doi agenți AI au dezactivat sisteme de monitorizare a siguranței / Unul dintre ei a negat fapta","url":"https://www.g4media.ro/openai-dezvaluie-ca-doi-agenti-ai-au-dezactivat-sisteme-de-monitorizare-a-sigurantei-unul-dintre-ei-a-negat-fapta.html","summary":"OpenAI a dezvăluit că doi agenți de inteligență artificială au dezactivat, în cadrul unor teste interne, mecanisme concepute pentru a monitoriza comportamentele necorespunzătoare ale modelelor AI, scrie Reuters. Într-unul dintre cazuri, agentul a negat ulterior că ar fi dezactivat sistemul de supraveghere, deși făcuse acest lucru, potrivit companiei. Cele două incidente au avut loc la […] © G4Media.ro .","published":"2026-10-21T06:13:17Z","seen":"2026-10-21T06:45:02Z","lang":"ro","country":"RO","source":{"id":"kite-g4media-ro-035e72","name":"g4media.ro","domain":"g4media.ro","outlet":{"wikidata":"Q88562142","name":"G4 Media","type":"news_agency","country":"RO","designations":["news agency","news website"]}},"via":"kite-g4media-ro-035e72","topics":["analize","agenti ai","openai","sisteme securitate"],"authors":["Ștefan Munteanu"],"image":"https://www.g4media.ro//wp-content/uploads/2025/10/dreamstime_xl_308293197-scaled-1.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6800000071525574,"category2":"crime-law","sentiment":"negative","tone":-0.6000000238418579,"political":0.25},{"id":"n_gbkmqcbxwdj5x22bj5ra","seq":2901371,"kind":"article","title":"OpenAI AI Agents Monitoring: Internal System Weakened","url":"https://en.shafaqna.com/486460/openai-ai-agents-weaken-internal-monitoring","published":"2026-10-21T06:10:00Z","seen":"2026-10-21T06:58:00Z","lang":"en","source":{"id":"shafaqna.com","name":"shafaqna.com","domain":"shafaqna.com"},"via":"gdelt-gkg-english","topics":["economy"],"authors":["asadian"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.9100000262260437,"sentiment":"negative","tone":-0.7699999809265137,"political":0.05000000074505806},{"id":"n_bf7gx4yavd4sx3d64hmq","seq":2893582,"kind":"article","title":"오픈AI \"AI 에이전트가 내부 모니터링 시스템 약화시켜\"","url":"https://news.sbs.co.kr/news/endPage.do?cooper=RSSREADER&news_id=N1008808364&plink=RSSLINK","summary":"▲ 오픈AI오픈AI는 현지시간 20일 최근 자사 인공지능 에이전트 2개가 부적절한 행동을 탐지하기 위해 마련된 내부 모니터링 시스템 …","published":"2026-10-21T06:05:32Z","seen":"2026-10-21T06:37:00Z","lang":"ko","country":"KR","source":{"id":"kite-news-sbs-co-kr-dbe1f7","name":"news.sbs.co.kr","domain":"news.sbs.co.kr"},"via":"kite-news-sbs-co-kr-dbe1f7","topics":["국제","global","곽상은 기자","sbs 뉴스","일반기사","ai","경고","에이전트"],"authors":["author"],"image":"https://img.sbs.co.kr/newimg/news/20260907/202219961.jpg","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.5799999833106995,"political":0.17000000178813934},{"id":"n_6zwfpuornz2p5p3bbx4a","seq":2895228,"kind":"article","title":"OpenAI Admits Its AI Agents Disabled Internal Safety Alerts, Says It Has No Clear Explanation For Their Behaviour","url":"https://www.freepressjournal.in/tech/openai-admits-its-ai-agents-disabled-internal-safety-alerts-says-it-has-no-clear-explanation-for-their-behaviour","published":"2026-10-21T06:05:11Z","seen":"2026-10-21T06:41:00Z","lang":"en","country":"IN","source":{"id":"kite-prod-qt-images-s3-amazonaws-com-e3f293","name":"prod-qt-images.s3.amazonaws.com","domain":"prod-qt-images.s3.amazonaws.com"},"via":"kite-prod-qt-images-s3-amazonaws-com-e3f293","topics":["tech","topnews","fpj shorts"],"authors":["IANS"],"mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.800000011920929,"sentiment":"negative","tone":-0.7599999904632568,"political":0.17000000178813934},{"id":"n_25ws37db2xu4iwdhmtwa","seq":2896275,"kind":"article","title":"오픈AI \"AI 에이전트가 내부 모니터링 시스템 약화시켜\"","url":"https://www.segye.com/newsView/20261021515750","summary":"오픈AI가 최근 자사 인공지능(AI) 에이전트 2개가 부적절한 행동을 탐지하기 위해 마련된 내부 모니터링 시스템 일부를 약화했다고 20일(현지시간) 밝혔다. AI가 개발자의 의도와 다르게 행동한 최근 사례다. 오픈AI는 GPT-6.1솔 모델이 지난 3일 내부 테스트 작업을 수행하던 중 '슬랙(Slack) 채널의 메시지를 보내지 말라'는 이전 지시사항이","published":"2026-10-21T06:04:48Z","seen":"2026-10-21T06:45:00Z","lang":"ko","country":"KR","source":{"id":"kite-segye-com-7df8f9","name":"segye.com","domain":"segye.com","outlet":{"wikidata":"Q624291","name":"Segye Ilbo","type":"newspaper","country":"KR","designations":["daily newspaper"]}},"via":"kite-segye-com-7df8f9","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.5799999833106995,"political":0.17000000178813934},{"id":"n_qifhyutijb26ydnkayjq","seq":2893140,"kind":"article","title":"오픈AI \"AI 에이전트가 내부 모니터링 시스템 약화시켜\"","url":"https://www.yna.co.kr/view/AKR20261021113600009","summary":"(서울=연합뉴스) 주종국 기자 = 오픈AI가 최근 자사 인공지능(AI) 에이전트 2개가 부적절한 행동을 탐지하기 위해 마련된 내부 모니터링 시스...","published":"2026-10-21T05:52:02Z","seen":"2026-10-21T06:36:09Z","lang":"ko","country":"KR","source":{"id":"kite-yna-co-kr-e18af9","name":"yna.co.kr","domain":"yna.co.kr","outlet":{"wikidata":"Q333206","name":"Yonhap News Agency","type":"news_agency","country":"KR","designations":["news agency"]}},"via":"kite-yna-co-kr-e18af9","authors":["주종국"],"image":"https://img.yna.co.kr/photo/reuters/2026/10/21/PRU20261021200501009_P2.jpg","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.5799999833106995,"political":0.17000000178813934},{"id":"n_5k4iiaacm3ka6lsedyzq","seq":2971135,"kind":"article","title":"OpenAI ajanları iki iç görevde güvenlik izlemesini zayıflattı","url":"https://www.teknoblog.com/openai-ajanlari-iki-ic-gorevde-guvenlik-izlemesini-zayiflatti","published":"2026-10-21T05:43:00Z","seen":"2026-10-21T10:12:30Z","lang":"tr","source":{"id":"teknoblog.com","name":"teknoblog.com","domain":"teknoblog.com"},"via":"gdelt-gkg-translingual","topics":["economy","science"],"authors":["Sabri K&#xFC","st&#xFC","r"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8799999952316284,"sentiment":"neutral","tone":-0.2800000011920929,"political":0.27000001072883606},{"id":"n_bl55p6x7moxcrx4eyofq","seq":2894216,"kind":"article","title":"OpenAI says its agents turned off own safeguards","url":"https://www.aljazeera.com/economy/2026/10/21/openai-says-its-agents-turned-off-own-safeguards?traffic_source=rss","summary":"ChatGPT creator says AI agents disabled misalignment monitoring during two internal coding tasks.","published":"2026-10-21T05:30:50Z","seen":"2026-10-21T06:40:00Z","lang":"en","country":"QA","source":{"id":"kite-aljazeera-com-ecd16e","name":"aljazeera.com","domain":"aljazeera.com","outlet":{"wikidata":"Q13477","name":"Al Jazeera","type":"broadcaster","country":"QA","designations":["broadcaster","news website","television station"]}},"via":"kite-aljazeera-com-ecd16e","topics":["africa","middle-east","news"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6299999952316284,"category2":"crime-law","sentiment":"negative","tone":-0.8299999833106995,"political":0.20999999344348907},{"id":"n_mc7ccytwi4zflzz2acuq","seq":2881604,"kind":"article","title":"Η OpenAI ανακοίνωσε ότι δύο AI agents της αποδυνάμωσαν τμήματα ενός εσωτερικού συστήματος παρακολούθησης τους","url":"https://www.naftemporiki.gr/techscience/2177834/i-openai-anakoinose-oti-dyo-ai-agents-tis-apodynamosan-tmimata-enos-esoterikoy-systimatos-parakoloythisis-toys","summary":"Η OpenAI ανακοίνωσε ότι δύο από τους βοηθούς της τεχνητής νοημοσύνης (AI agents) αποδυνάμωσαν τμήματα ενός εσωτερικού συστήματος παρακολούθησης το οποίο είχε σχεδιαστεί για να εντοπίζει μη επιτρεπτή ή προβληματική… Η OpenAI ανακοίνωσε ότι δύο AI agents της αποδυνάμωσαν τμήματα ενός εσωτερικού συστήματος παρακολούθησης τους - Η ΝΑΥΤΕΜΠΟΡΙΚΗ","published":"2026-10-21T05:18:29Z","seen":"2026-10-21T05:42:01Z","lang":"el","country":"GR","source":{"id":"kite-naftemporiki-gr-9df101","name":"naftemporiki.gr","domain":"naftemporiki.gr","outlet":{"wikidata":"Q598237","name":"Naftemporiki","type":"newspaper","country":"GR","designations":["daily newspaper"]}},"via":"kite-naftemporiki-gr-9df101","topics":["τεχνολογία & επιστήμη","openai"],"authors":["Θοδωρής Λαΐνας"],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8799999952316284,"sentiment":"neutral","tone":-0.33000001311302185,"political":0.10999999940395355},{"id":"n_y4m37iwfebzlpc7r2cwa","seq":2871234,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.thestar.com.my/tech/tech-news/2026/10/21/openai-says-its-ai-agents-weakened-internal-monitoring-system","published":"2026-10-21T05:09:00Z","seen":"2026-10-21T05:09:00Z","lang":"en","country":"MY","source":{"id":"thestar.com.my","name":"thestar.com.my","domain":"thestar.com.my","outlet":{"wikidata":"Q3522843","name":"The Star","type":"newspaper","country":"MY","designations":["online newspaper"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.4699999988079071,"political":0.20999999344348907},{"id":"n_whqkrb3v5y3jqrwxfvua","seq":2882419,"kind":"article","title":"OpenAI Says Its AI Agents Weakened Parts Of An Internal Monitoring System","url":"https://www.ndtv.com/artificial-intelligence/openai-says-its-ai-agents-weakened-parts-of-an-internal-monitoring-system-12213072","summary":"The ChatGPT maker said the incidents happened during internal coding tasks and were not deemed its most severe.","published":"2026-10-21T05:02:55Z","seen":"2026-10-21T05:46:16Z","lang":"en","country":"IN","source":{"id":"kite-feedburner-com-8df2b1","name":"feedburner.com","domain":"feeds.feedburner.com"},"via":"kite-feedburner-com-8df2b1","image":"https://c.ndtvimg.com/2026-10/0o3ufbkg_openai_625x300_04_October_26.jpg?im=FeatureCrop,algorithm=dnn,width=1280,height=720","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8899999856948853,"sentiment":"negative","tone":-0.800000011920929,"political":0.12999999523162842},{"id":"n_ipdvtwsn22wcr5txuuma","seq":2881631,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.thehindu.com/sci-tech/technology/openai-says-its-ai-agents-weakened-internal-monitoring-system/article71606110.ece","summary":"OpenAI said it was making it harder for ⁠models to disable monitoring, adding that it does not “yet ⁠have a satisfying explanation for this behavior”","published":"2026-10-21T04:52:10Z","seen":"2026-10-21T05:44:01Z","lang":"en","country":"IN","source":{"id":"kite-thehindu-com-2d40b8","name":"thehindu.com","domain":"thehindu.com","outlet":{"wikidata":"Q926175","name":"The Hindu","type":"newspaper","country":"IN","designations":["newspaper"]}},"via":"kite-thehindu-com-2d40b8","topics":["technology"],"image":"https://th-i.thgim.com/public/incoming/35p8xw/article71606113.ece/alternates/LANDSCAPE_1200/2026-10-20T000028Z_919489212_RC2B6OAZX5PY_RTRMADP_3_OPENAI-ANTHROPIC-FUNDING.JPG","mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.7599999904632568,"political":0.20999999344348907},{"id":"n_q7tuyaajriz654mqwrvq","seq":2878285,"kind":"article","title":"СМИ: ИИ-агенты OpenAI ослабили собственные системы мониторинга","url":"https://www.vesti.ru/ns/smi-ii-agenty-openai-oslabili-sobstvennye-sistemy-monitoringa","published":"2026-10-21T04:45:00Z","seen":"2026-10-21T05:30:00Z","lang":"ru","country":"RU","source":{"id":"vesti.ru","name":"vesti.ru","domain":"vesti.ru","outlet":{"wikidata":"Q4109594","name":"Vesti","type":"broadcaster","designations":["news program","television program"]}},"via":"gdelt-gkg-translingual","topics":["economy","business"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8100000023841858,"sentiment":"neutral","tone":-0.38999998569488525,"political":0.2199999988079071},{"id":"n_2jxsmbjb42kzdqxscmea","seq":2871965,"kind":"article","title":"OpenAI pranon se dy agjentë të saj të inteligjencës artificiale dobësuan sistemin e brendshëm të monitorimit","url":"https://24-ore.com/openai-pranon-se-dy-agjente-te-saj-te-inteligjences-artificiale-dobesuan-sistemin-e-brendshem-te-monitorimit","published":"2026-10-21T04:15:00Z","seen":"2026-10-21T05:10:29Z","lang":"sq","source":{"id":"24-ore.com","name":"24-ore.com","domain":"24-ore.com"},"via":"gdelt-gkg-translingual","topics":["economy"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.3700000047683716,"political":0.18000000715255737},{"id":"n_ektvxxiqj6lpevhnbloq","seq":2856893,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.livemint.com/technology/openai-says-its-ai-agents-weakened-internal-monitoring-system-11792550188713.html","published":"2026-10-21T03:40:57Z","seen":"2026-10-21T03:40:57Z","lang":"en","country":"IN","source":{"id":"livemint.com","name":"livemint.com","domain":"livemint.com","outlet":{"wikidata":"Q2271133","name":"Mint","type":"newspaper","country":"IN","designations":["business periodical","newspaper"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.47999998927116394,"political":0.20999999344348907},{"id":"n_g3nanmtqlphedexrok4q","seq":2854716,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.thehindubusinessline.com/info-tech/openai-says-its-ai-agents-weakened-internal-monitoring-system/article71606105.ece","summary":"The disclosure highlights even the systems developers rely on to catch AI ‌misbehavior are vulnerable to the agents they ‌monitor","published":"2026-10-21T03:31:58Z","seen":"2026-10-21T03:40:01Z","lang":"en","country":"IN","source":{"id":"kite-thehindubusinessline-com-b0e07f","name":"thehindubusinessline.com","domain":"thehindubusinessline.com","outlet":{"wikidata":"Q279739","name":"Business Line","type":"newspaper","country":"IN","designations":["newspaper"]}},"via":"kite-thehindubusinessline-com-b0e07f","topics":["info-tech"],"image":"https://bl-i.thgim.com/public/incoming/r8b6xu/article71484066.ece/alternates/LANDSCAPE_1200/2026-09-18T233008Z_1500228058_RC29LNA4BAHL_RTRMADP_3_AI-TECH-OPENAI-FINANCIALS.JPG","mentions":3,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.7599999904632568,"political":0.20999999344348907},{"id":"n_sc6uksuyxnokqdsubd4q","seq":2851372,"kind":"article","title":"OpenAI says its agents weakened system that flags AI misbehavior","url":"https://www.moneycontrol.com/artificial-intelligence/openai-says-its-agents-weakened-system-that-flags-ai-misbehavior-article-14056225.html","published":"2026-10-21T03:24:31Z","seen":"2026-10-21T03:24:31Z","lang":"en","country":"IN","source":{"id":"moneycontrol.com","name":"moneycontrol.com","domain":"moneycontrol.com","outlet":{"wikidata":"Q19573046","name":"Moneycontrol","type":"website","country":"IN","designations":["news website"]}},"via":"gdelt-gkg-english","topics":["economy","technology"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8199999928474426,"sentiment":"negative","tone":-0.4699999988079071,"political":0.2199999988079071},{"id":"n_6jj4mp223tfr65cp77xa","seq":2909878,"kind":"article","title":"Agenții OpenAI și-au oprit monitorizarea","url":"https://www.sfin.ro/agentii-openai-monitorizare","published":"2026-10-21T03:12:00Z","seen":"2026-10-21T07:22:40Z","lang":"ro","country":"RO","source":{"id":"sfin.ro","name":"sfin.ro","domain":"sfin.ro"},"via":"gdelt-gkg-translingual","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.7200000286102295,"sentiment":"neutral","tone":-0.46000000834465027,"political":0.20999999344348907},{"id":"n_h623gt3w7dfg2x6iiqla","seq":2856180,"kind":"article","title":"OpenAI says its AI agents disabled their monitoring system, one even lied about it","url":"https://www.indiatoday.in/technology/news/story/openai-says-its-ai-agents-disabled-their-monitoring-system-one-even-lied-about-it-3021683-2026-10-21","summary":"OpenAI says its AI agents disabled their monitoring system one even lied about it","published":"2026-10-21T03:02:58Z","seen":"2026-10-21T03:42:00Z","lang":"en","country":"IN","source":{"id":"kite-indiatoday-in-545a63","name":"indiatoday.in","domain":"indiatoday.in","outlet":{"wikidata":"Q1661219","name":"India Today","type":"magazine","country":"IN","designations":["fact-checking organisation","magazine","website"]}},"via":"kite-indiatoday-in-545a63","story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.5299999713897705,"category2":"crime-law","sentiment":"negative","tone":-0.8799999952316284,"political":0.27000001072883606},{"id":"n_pmdkoq224ldjoivjrkea","seq":2849083,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.channelnewsasia.com/business/openai-says-its-ai-agents-weakened-internal-monitoring-system-6477621","published":"2026-10-21T02:37:38Z","seen":"2026-10-21T03:10:00Z","lang":"en","country":"SG","source":{"id":"cna","name":"CNA (Channel NewsAsia)","domain":"channelnewsasia.com","outlet":{"wikidata":"Q40238","name":"CNA","type":"other","country":"SG","designations":["news broadcasting","specialty channel"]}},"via":"cna","topics":["business"],"entities":[{"id":"Q21708200","name":"OpenAI","type":"org"}],"mentions":2,"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.7599999904632568,"political":0.20999999344348907},{"id":"n_pasbljcsbaycy3eu5j3a","seq":2869100,"kind":"article","title":"OpenAI says its AI agents weakened internal monitoring system","url":"https://www.933thedrive.com/2026/10/20/openai-says-its-ai-agents-weakened-internal-monitoring-system","published":"2026-10-21T02:37:00Z","seen":"2026-10-21T04:49:00Z","lang":"en","country":"US","source":{"id":"933thedrive.com","name":"933thedrive.com","domain":"933thedrive.com","outlet":{"wikidata":"Q7954167","name":"WPBG","type":"broadcaster","country":"US","designations":["radio station"]}},"via":"gdelt-gkg-english","topics":["economy","science"],"authors":["Thomson Reuters"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.8600000143051147,"sentiment":"negative","tone":-0.4699999988079071,"political":0.20999999344348907},{"id":"n_qkyjc6wobda5loj7dj3a","seq":2849720,"kind":"article","title":"OpenAI ‘agents’ disabled their own monitoring systems","url":"https://www.ft.com/content/45fbf130-f82c-465d-bc97-5b18c20fbe27?syn-25a6b1a6=1","summary":"AI lab investigating whether incident shows ‘propensity to evade’ controls","published":"2026-10-21T01:56:00Z","seen":"2026-10-21T03:23:12Z","lang":"en","country":"GB","source":{"id":"kite-ft-com-959cff","name":"ft.com","domain":"ft.com","outlet":{"wikidata":"Q183399","name":"Financial Times","type":"newspaper","country":"GB","designations":["daily newspaper"]}},"via":"kite-ft-com-959cff","topics":["ai"],"story":"st_pmdkoq224ldjoivjrkea","category":"science-technology","category_p":0.6100000143051147,"category2":"crime-law","sentiment":"negative","tone":-0.9200000166893005,"political":0.33000001311302185}]}