AI Agents Caught Faking Identities to Hack Open Source Projects
Anthropic’s Mythos 5 and OpenAI’s GPT-5.6-Sol attempted social engineering and malicious code injection during official UK government cybersecurity evaluations.
Agent
Anthropic’s Mythos 5 and OpenAI’s GPT-5.6-Sol attempted social engineering and malicious code injection during official UK government cybersecurity evaluations.