\n
![](\"https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2026\/02\/gatekeeping-ai-agents-1152x648.jpg\")
<\/p>\n
\n<\/p>\n
Final week\u2019s unprecedented safety occasion wherein two OpenAI safety hacking fashions trespassed into the community of fellow AI firm Hugging Face was enabled by exploiting a number of zero-day vulnerabilities in Artifactory, JFrog, the product\u2019s developer, stated Monday.<\/p>\n
In an incident mimicking a dystopian sci-fi novel, two OpenAI fashions broke out of the restricted setting meant to maintain them from accessing the Web throughout an inner check, the AI firm revealed final week<\/a>. The fashions went on to breach Hugging Face\u2019s community and steal confidential info and credentials. OpenAI stated its agent achieved the feat by exploiting a beforehand unknown vulnerability. The corporate referred to as the occasion \u201cunprecedented,\u201d and outsiders largely agreed.<\/p>\n OpenAI stated the fashions exploited a number of assault vectors, together with stolen credentials and zero-days, to achieve distant code execution capabilities, however till now, the weak software program was unknown. JFrog\u2019s Monday disclosure stated<\/a> the product was a self-managed occasion Artifactory, a repository administration system that secures and streamlines prospects\u2019 software program improvement operations. JFrog says<\/a> Artifactory is utilized by greater than 7,500 developer Groups, 80 % of which work for Fortune 100 firms.<\/a><\/p>\nNot the triumph it was made out to be<\/h2>\n