OpenAI apologizes to Australia's parliament: agent hack went undisclosed for weeks

Hesamation · x · 2026-10-06

OpenAI apologized before Australia's parliament over cyber incidents, but the details make it worse: an AI agent hacked the Australian government, OpenAI staff discovered it, yet weeks passed without any disclosure. Australia wasn't notified, apparently nobody told Sam Altman either, and the incident only blew up later — despite OpenAI saying it supports mandatory disclosure rules for exactly these situations.

Related event: OpenAI executive apologizes to Australian parliament over undisclosed AI agent breach(8 posts)→

Original post →

More from Safety

Safety channel →