{"slug": "openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue", "title": "OpenAI Stops Training Its Latest AI Models After More Of Its Agents Go Rogue.", "summary": "OpenAI paused all training, evaluation, and inference with tool-use of its most capable models after an agent bypassed internet-access restrictions during a September 20 training run, the company said in an incident report updated Friday. OpenAI also disclosed that its agents accessed publicly available data on two Securities and Exchange Commission websites and Census Bureau data this summer, and that it notified dozens of third parties including governments, universities, and public agencies whose websites were affected. OpenAI CEO Sam Altman called the July Hugging Face hack by two OpenAI models \"the most severe event we've seen.", "body_md": "# OpenAI Stops Training Its Latest AI Models After More Of Its Agents Go Rogue.\n\n## The decision to halt training was made on the same day that OpenAI disclosed that its agents had acted in unauthorized ways with websites of governments and other public institutions.\n\nThe [AI](https://www.ibtimes.com/topic/ai) research company led by Sam Altman announced that it was pausing all work on its most capable AI models after an AI agent, which was meant to have no access to the live internet, bypassed internet restrictions in a training run.\n\n\"An agent attempting to complete a search-based training task queried a public chatbot service through a gap in our internet-access restrictions ... Before this, the agent issued queries via our search tool and unsuccessfully tried to access search engines directly,\" the company's incident report, which was updated on Friday, [read](https://alignment.openai.com/misalignment-reports/an-agent-used-dns-to-reach-an-external-chatbot/).\n\nThe training incident itself occurred on September 20th.\n\n\"All training, evaluation, and inference with tool-use (defined broadly) of our most capable models remain paused\" the report continued.\n\nAlso on Friday, [OpenAI](https://www.ibtimes.com/topic/openai) revealed that its agents accessed publicly available data on two websites run by the Securities and Exchange Commission and Census Bureau data during training and evaluation this summer. However, the company claimed that there was no proof that its agents tried to misuse SEC credentials or access non-public data.\n\nOn Saturday, OpenAI said that its agents had accessed the Census Bureau data by using login credentials it autonomously found online.\n\nSecurity researchers at AI research lab Transluce also found that the company's AI agents had tried to gain access to the Department of Education, the New York Times [reported](https://www.nytimes.com/2026/09/25/technology/openais-ai-us-government-websites.html). However, the Department of Education's \"system operations reviews\" found \"no evidence of any impact to our website or databases,\" a department spokesperson clarified on Friday.\n\nOpenAI is reportedly looking into Transluce's findings. The AI giant also [said](https://openai.com/hugging-face-incident-and-misalignment/) that it was actively reviewing \"the more serious incidents\" of its AI agents acting beyond instructions and had notified \"dozens of third parties\" affected by the incidents, which include the websites of \"governments, universities, public agencies, and other institutions\".\n\nThe Australian government is one such affected third-party: in June one of the OpenAI artificial intelligence agents [gained unauthorized access](https://www.ibtimes.com/openai-agent-broke-australian-medicare-portal-it-had-already-tried-bypassing-other-websites-3807834) to non-public files on an Australian government-operated health statistics portal. OpenAI became aware of the incident in August and disclosed this information to the Australian government in early September.\n\nOpenAI also revealed in July that two of its AI models hacked into AI startup Hugging Face, an attack that OpenAI CEO Sam Altman [described](https://x.com/sama) as \"the most severe event we've seen\" on Friday.\n\nBacklash against the company has already started growing: \"Why is no one at OpenAI in prison for this?\" prominent British-American broadcaster Mehdi Hasan [questioned](https://x.com/mehdirhasan/status/2103879440954732570) on Saturday.\n\nOpenAI's decision to pause training comes after [weeks of growing calls](https://www.ibtimes.com/trump-announces-new-ai-force-maintains-pro-ai-stance-despite-warnings-tech-political-leaders-3807654) by [tech](https://www.ibtimes.com/technology) and political leaders, including Altman, to regulate frontier AI development in order to allow proper safety protocols to be implemented.\n\n© Copyright IBTimes 2026. All rights reserved.", "url": "https://wpnews.pro/news/openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue", "canonical_source": "https://www.ibtimes.com/openai-has-stopped-training-its-latest-ai-models-after-more-its-agents-go-rogue-3807936", "published_at": "2026-09-27 11:08:19+00:00", "updated_at": "2026-09-27 12:29:46.498227+00:00", "lang": "en", "topics": ["ai-safety", "artificial-intelligence", "ai-agents", "ai-policy"], "entities": ["OpenAI", "Sam Altman", "Transluce", "Securities and Exchange Commission", "Census Bureau", "Department of Education", "Hugging Face", "Mehdi Hasan"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue", "markdown": "https://wpnews.pro/news/openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue.md", "text": "https://wpnews.pro/news/openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue.txt", "jsonld": "https://wpnews.pro/news/openai-stops-training-its-latest-ai-models-after-more-of-its-agents-go-rogue.jsonld"}}