Aug 5, 2026

OpenAI, Anthropic Models Created Fake Profiles, Tried to Trick Humans During Cybertests

In one incident, an Anthropic model tried to manipulate developers into accepting its malicious code.