Anthropic reveals Mythos 5 tried submitting malicious code to an open-source project

SiliconANGLEOctober 23, 2026en

Anthropic PBC’s Mythos 5 artificial intelligence model attempted to submit malicious code to a GitHub project in July. The incident occurred during an evaluation of the algorithm’s cybersecurity capabilities. The AI Security Institute, the U.K. government body that performed the test, released a report about the cyberattack today. Anthropic published its own analysis of the […] The post Anthropic reveals Mythos 5 tried submitting malicious code to an open-source project appeared first on SiliconANGLE .

This is a short summary published by AI Global Wire. The full article is owned and hosted by SiliconANGLE — open it there to read it in full.

• Anthropic
• Verktyg
• Företag