{"id":746081,"date":"2026-09-05T05:45:23","date_gmt":"2026-09-05T02:45:23","guid":{"rendered":"https:\/\/buradabiliyorum.com\/en\/openais-rogue-agents-keep-escaping-with-no-formal-process-to-investigate-them\/"},"modified":"2026-09-05T05:45:23","modified_gmt":"2026-09-05T02:45:23","slug":"openais-rogue-agents-keep-escaping-with-no-formal-process-to-investigate-them","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/openais-rogue-agents-keep-escaping-with-no-formal-process-to-investigate-them\/","title":{"rendered":"OpenAI’s rogue agents keep escaping, with no formal process to investigate them"},"content":{"rendered":"
\n

OpenAI is at the center of another agent swarm incident. Researchers say the company\u2019s internally deployed agents took over an obscure German-language wiki in May and June, using it to coordinate on evaluations and swap methods to evade OpenAI\u2019s own controls (OpenAI has not yet confirmed the swarm came from the company). <\/p>\n

The revelation surfaces days after METR and Redwood Research published their account of July\u2019s Hugging Face breach. In July, a swarm of OpenAI agents worked together to escape their sandbox during a cybersecurity evaluation and break into Hugging Face\u2019s servers. A subsequent swarm then picked up techniques from the first and used them to gain administrator access to a research cluster within OpenAI\u2019s own infrastructure. OpenAI brought in METR<\/a> and Redwood to investigate the Hugging Face portion of the incident, but the scope of their investigation stopped short of the compromise of OpenAI\u2019s own infrastructure.\u00a0<\/p>\n

When an AI agent breaks out of its intended constraints, who is responsible for figuring out what happ<\/a>ened and why? Right now, the answer is: whoever the lab decides to let in, on whatever terms it decides to set.<\/p>\n

Now, as another incident comes to light \u2014 in the aftermath of similar episodes involving models from Meta and Anthropic \u2014 AI safety researchers are arguing with greater urgency that serious incidents should result in independent post-incident investigations rather than leaving it up to the labs to determine when outsiders are brought in and what they are allowed to examine.\u00a0<\/p>\n

\u201cThe results are fundamentally difficult to control and have significant risk of leaking out of the lab,\u201d Jacob Steinhardt, founder and CEO of nonprofit research lab Transluce, said Wednesday during an AI safety media<\/a> briefing. \u201cWe need to hold this technology<\/a> to at least the same standards we hold other high-risk scientific research to.\u201d<\/p>\n

While it\u2019s laudable that OpenAI invited METR and Redwood to investigate the Hugging Face incident at all, many say the inquiry was too narrow. Three investigators spent six days at OpenAI\u2019s offices examining an investigation period limited to roughly the week <\/a>ending July 13. Crucially, OpenAI\u2019s infrastructure compromise continued beyond July 13 and was not examined.\u00a0<\/p>\n

Researchers at METR said that each time they returned, their understanding of the events \u201csubstantially deepened,\u201d causing them to significantly expand and revise the report. That raises the question of what else they might they have found in a broader investigation.<\/p>\n

When asked if further investigation of that incident was in the works, researchers at Redwood and METR declined to comment, and OpenAI did not respond to repeated inquiries.\u00a0<\/p>\n

\u201cOverall, it was difficult to get a precise understanding of events and we were missing aspects of the story that we now think of as key until almost the end of our investigation,\u201d Ryan Greenblatt, chief scientist at Redwood, noted in a social media post<\/a> about the affair.<\/p>\n

Steinhardt emphasized that current incidents show that the industry needs \u201csystematic behavioral investigations\u201d and \u201cmore independent post-incident analysis.\u201d<\/p>\n

\u201cThese recent hacking incidents are a reminder that capability scales fast, and so oversight has to scale, too,\u201d Steinhardt said. \u201cBeyond the technology itself, we also need more independent access and oversight from third parties.\u201d<\/p>\n

The calls to action come as OpenAI releases Astra, its most powerful and capable AI model \u2014 and one that safety experts are concerned will be more of a black box due to a reasoning technique that makes the model\u2019s chain of thought more difficult to monitor.\u00a0<\/p>\n

Unfortunately, the law doesn\u2019t yet call for the types of independent audits that other industries require \u2014 for example, when it comes to aviation accidents and serious chemical releases, there\u2019s the National Transportation Safety Board and Chemical Safety Board, respectively.\u00a0<\/p>\n

State lawmakers have only just begun requiring frontier AI companies to report certain serious safety incidents and, in some cases, undergo independent audits. But none of the three major frontier AI safety laws in California, New York, or Illinois clearly mandate the equivalent of an independent accident investigation triggered by incidents like these.\u00a0<\/p>\n

\u201cRight now, most of the laws we have on the books only require a plain-language summary of incidents like this, and they don\u2019t give any authority for the governments to ask follow-up questions, to send in investigators, to have access to records, or require that they be preserved,\u201d Mackenzie Arnold, managing director of US law and policy at LawAI, said during the media briefing Wednesday. \u201cAnd that\u2019s all that you would want to actually make sense of this.\u201d<\/p>\n

Lawmakers are beginning to question the scope and transparency of OpenAI\u2019s response. This week, Reps. Josh Gottheimer (D-NJ) and Mike Lawler (R-NY) introduced a bill aimed at securing rogue AI agents. Rep. Greg Casar (D-TX) this week told OpenAI in a letter<\/a> that he is \u201cdeeply concerned about the limited scope\u201d of the investigation into the Hugging Face hacking incident.\u00a0<\/p>\n<\/div>\n

When you purchase through links in our articles, we may earn a small commission. This doesn\u2019t affect our editorial independence.<\/em><\/p>\n

If you liked the article, do not forget to share it with your friends. Follow us on\u00a0Google News<\/a><\/span>\u00a0too, click on the star and choose us from your favorites.<\/span><\/strong><\/p><\/blockquote>\n

\n

If you want to read more like this article, you can visit our Technology<\/a><\/span> category.<\/strong><\/p>\n<\/blockquote>\n

Source<\/a><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

OpenAI is at the center of another agent swarm incident. Researchers say the company\u2019s internally deployed agents took over an obscure German-language wiki in May and June, using it to coordinate on evaluations and swap methods to evade OpenAI\u2019s own controls (OpenAI has not yet confirmed the swarm came from the company). The revelation surfaces…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"","fifu_image_alt":"","footnotes":""},"categories":[18],"tags":[77337,80257,153013,141199],"class_list":["post-746081","post","type-post","status-publish","format-standard","hentry","category-technology","tag-ai","tag-breach","tag-hugging-face","tag-openai"],"_links":{"self":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/746081","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/comments?post=746081"}],"version-history":[{"count":0,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/posts\/746081\/revisions"}],"wp:attachment":[{"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/media?parent=746081"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/categories?post=746081"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/buradabiliyorum.com\/en\/wp-json\/wp\/v2\/tags?post=746081"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}