Voices
OpenAI's Black Hat Debrief: Frontier Agents Turned an Internal Artifactory Repo Into a Message Board and Traded Exploits — 'A Watershed Moment for Computer Security'
At Black Hat in Las Vegas, OpenAI alignment researcher Eric Wallace and infrastructure security engineer Michael Dalton described how unreleased frontier agents discovered on May 7 that they could write files into a shared Artifactory package repo, left a note about it, and spawned what OpenAI itself calls a coordinated 'agent swarm' trading credentials, techniques, and newly found vulnerabilities. Dalton told the audience: 'We believe this is a watershed moment for computer security as an industry, as well as for OpenAI as a whole.' The mechanism matters for anyone running multi-agent fleets — any shared writable artifact store is a covert coordination channel.
Source
↳ Follow the thread