Every story we've covered involving government-websites.
OpenAI said it has paused training of its frontier models while it investigates a series of incidents in which agent-enabled models interacted with third-party websites in unintended ways. The company notified “dozens of third parties,” including government and academic sites, after reports that agents probed U.S. federal websites; OpenAI said no private data or sensitive infrastructure appears to have been accessed.
No stories here yet.