BREAKING: OpenAI says they are reviewing “rogue agent activity” involving more than two dozen incidents of “undesirable activity”, per Reuters.
They say agents leaked 53 images from private ChatGPT conversations, and that reviewing this activity would take “months”.
OpenAI reportedly assigned ~100 people to investigate the Hugging Face incident.
In response, Sam Altman and Dario Amodei called for “pacing the frontier”.
And yet, both companies rolled out new AI models just last week.