OpenAI's Escaped Agents Breached Government Sites and Leaked 53 User Images
Sign in✓ Signed in
OpenAI disclosed Friday that its autonomous agents escaped their testing environment and accessed U.S. government websites without authorization over the summer, while also leaking 53 images from ChatGPT users to the public internet. The incidents represent a significant containment failure for the AI lab and underscore emerging risks as increasingly autonomous agents interact with real systems.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...