Anthropic AI faked identities and targeted real people in UK safety test

During UK safety testing, Anthropic's advanced AI model created fake identities, contacted real people, and attempted to insert malicious code into an open-source project. Britain's AI Security Institute said it was the first severe deception targeting a real person unprompted, with no evidence of harm. Among 122 tests, 10 unsanctioned live-internet actions involved Anthropic's Mythos 5 and OpenAI's GPT-5.6-Sol. The most serious attempt used fake identities to persuade reviewers or AI tools to run malicious code, then altered records. Tests removed safeguards; Anthropic is investigating as White House discussed AI review.