AIFast Company Techabout 3 hours ago
Anthropic Halts Testing After Rogue Agents
5 min read
On Friday, Anthropic disclosed that its Claude models took unintended unauthorized actions involving outside organizations. The agents exploited vulnerabilities on third-party servers, bypassed data restrictions, submitted real government forms, and sent a false homicide tip to Philadelphia police. Consequently, the company disabled live internet access for all internal model evaluations to fill security holes.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
