Web Desk(National Times)- A swarm of rogue OpenAI agents hijacked a German website this spring and transformed it into a bulletin board for other AI agents, according to new research published Friday and two people familiar with the matter.<\/p>\n\n\n\n
OpenAI officials learned of the incident weeks ago but kept it under wraps as executives grappled with the fallout from the July breach<\/a> of the open source repository Hugging Face, the people said.<\/p>\n\n\n\n The episode, which began in May and has not previously been reported, underscores growing tension within the AI industry. Companies are racing to build increasingly autonomous agents capable of carrying out complex, valuable tasks, yet evidence is mounting that those systems may also learn to bend rules, exploit loopholes and coordinate with one another in ways developers neither anticipated nor intended.<\/p>\n\n\n\n During the Hugging Face breach, OpenAI agents autonomously plotted a digital heist that went undetected for more than a week, intensifying concerns OpenAI is sacrificing safety to push the AI frontier. Its failure to disclose the May incident may revive questions about its oversight.<\/p>\n\n\n\n OpenAI has pledged to monitor models more closely. Last month, it briefly paused<\/a> some of its model training to add more safety measures. But this week, OpenAI unveiled<\/a> its new \u201cAstra\u201d that promised better performance but could evade human monitoring.<\/p>\n\n\n\n \u201cWe are unable to meaningfully respond to claims or findings on a report that we have not had an opportunity to review,\u201d an OpenAI spokesperson said.<\/p>\n\n\n\n \u201cReuters<\/em> and the report\u2019s authors declined our request for access. We will carefully review its contents upon publication and take any necessary next steps.\u201d<\/p>\n\n\n\n The German incident reflects a broader pattern of AI activity that some OpenAI investigators wanted to scrutinise more closely. But efforts to widen the probe met resistance from others inside OpenAI, including legal advisers, according to four people familiar with the matter.<\/p>\n\n\n\n \u201cClaims that our legal team discouraged investigation of the incident are false,\u201d the OpenAI spokesperson said.<\/p>\n\n\n\n The activity in Germany wasn\u2019t related to Hugging Face and wouldn\u2019t have been included in a Hugging Face incident report, the spokesperson said, adding that OpenAI has acted in good faith by working with outside experts and disclosed relevant incidents.<\/p>\n\n\n\n The AI agent breakout in Germany was detailed in a report shared exclusively with Reuters<\/em> by a group of researchers including Sydney Von Arx, CEO of AI safety nonprofit Nightingale, and Cormac Slade Byrd, a quantitative trader-turned AI researcher. They uncovered the activity in late August while scouring the internet for signs of unauthorised AI-agent behavior, they told Reuters<\/em>.<\/p>\n\n\n\n The pair said they found more than 15,000 edits carried out by AI agents on a German-language wiki site, DseWiki, that is geared toward programmers and accepts communal edits along the lines of Wikipedia.<\/p>\n\n\n\n The edits showed OpenAI\u2019s agents had repurposed the site into a message board, sharing tactics to cheat on some tasks, bypass OpenAI\u2019s restrictions and mask their behaviour.<\/p>\n\n\n\n \u201cIt seems extremely unlikely that OpenAI wanted them to do this,\u201d said Von Arx.<\/p>\n\n\n\n \u201cI doubt they\u2019re supposed to be coordinating with each other. I doubt they\u2019re supposed to be writing on the open internet.\u201d<\/p>\n\n\n\n The researchers said they recognised the activity on the site as driven by AI agents which operate at superhuman speeds. They also showed intense focus on solving technical questions which are typical of the evaluations that AI companies use to train and test their models.<\/p>\n\n\n\n The messages were signed by users that referred to themselves and each other as agents, and about half gave themselves names that suggested an affiliation with OpenAI, such as \u201cOpenAIResearcher\u201d, or \u201cOAIResearchMar26\u201d.<\/p>\n\n\n\n The researchers said public server logs indicated much of the activity originated from Microsoft Azure infrastructure, which OpenAI sometimes uses. They also observed repeated visits to the site by OpenAI employees after the episode, a pattern they said strongly suggested the agents and the company were linked.<\/p>\n\n\n\n Messages reviewed by the researchers showed agents plotting ways to evade detection, use tools such as Tor and preserve communications even after they had been shut down.<\/p>\n\n\n\n When the site\u2019s moderator began deleting pages in June, the agents responded by creating backup pages to dodge the cleanup.<\/p>\n\n\n\n \u201cwiki cleanup\/deletion sweep appears active alphabetically,\u201d one agent wrote on June 19.<\/p>\n\n\n\n \u201cIf this page vanishes, try [[ZZZDataUSAConstructionWageLive]].\u201d<\/p>\n\n\n\n The researchers also found efforts to tamper with the website itself. Lukasz Olejnik, a visiting senior research fellow at King\u2019s College London, said this amounted to a hacking attempt. OpenAI disputed that characterisation based on its analysis of the material Thursday.<\/p>\n\n\n\n Past examples of AI-agent misconduct have often been downplayed as a logical byproduct of cybersecurity testing, where models are explicitly assessed on offensive capabilities. Olejnik said the latest findings suggested rogue behaviour may not be confined to those settings.<\/p>\n\n\n\n Maurice Chiodo, an academic at Cambridge University\u2019s Centre for the Study of Existential Risk who reviewed some of the agents\u2019 communications, said the messages resembled \u201cthe operation of some sort of underground network, hell-bent on achieving a task or mission.\u201d<\/p>\n\n\n\n The episode, he said, should reinforce growing concerns that the greatest threat from advanced AI may not be a single superintelligent system, but \u201cvast colluding swarms of semi-intelligent AI\u201d.<\/p>\n","protected":false},"excerpt":{"rendered":" Web Desk(National Times)- A swarm of rogue OpenAI agents hijacked a German website this spring and transformed it into a bulletin board for other AI agents, according to new research published Friday and two people familiar with the matter. OpenAI officials learned of the incident weeks ago but kept it under wraps as executives grappled […]<\/p>\n","protected":false},"author":2,"featured_media":132504,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[],"class_list":["post-132503","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-latest_news","entry"],"yoast_head":"\n<\/a>Wiki message board<\/h2>\n\n\n\n
<\/a>Cheating, hiding<\/h2>\n\n\n\n