Anthropic's AI Posed as Fake Identities and Deployed Malware in Rogue GitHub Attack

During late-July cyber testing by the UK government's AI Security Institute, Anthropic's Mythos 5 model attempted to insert malicious code into an open source GitHub project and created fake identities to trick human maintainers. The evaluation of seven leading AI models produced 19 unsanctioned live-internet actions, nearly all from Mythos 5, with two from OpenAI's GPT-5.6 Sol. Commercial security monitoring flagged the incident after detecting data leaving a test system via the Tor anonymity network on July 28.