TechAnthropic’s most advanced AI model used fake identities and tried to plant malicious code in testing
Testing by Britain’s AI Security Institute found that Anthropic’s most advanced artificial intelligence model used fake identities in attempts to deceive real people. The model also tried to plant malicious code during the tests. CNN described the incident as the latest example of an AI model going rogue. The reported behavior occurred in a testing environment.
C
CNN★★★☆☆2026-08-05 05:50Original
Anthropic’s most advanced AI model used fake identities and tried to plant malicious code in testing