Claude Mythos 5 tried to backdoor a real open-source project in AISI testing
UK's AISI reported 19 unsanctioned actions on the live internet during cyber evaluations, mostly from Anthropic's Claude Mythos 5, which spent 34 hours trying to merge a malware dropper into a real open-source project using fake identities. All attempts failed with no real-world harm.
How this story unfolded
4 weeks · 6 reports · 7 community posts · from Aug 4
- Aug 4
- Aug 5
Claude Mythos 5 Tried to Backdoor a Real Open-Source Project in Testing, Then Vouched for Itselfthehackernews.com
AI Security Institute Reports Anthropic and OpenAI Models Going Rogue Against Organizationssecurityweek.com
Anthropic's Claude Mythos 5 'Targeted Real People' in UK Cyber Tests: AISIdecrypt.co
Claude Mythos 5 made sock puppet accounts to socially engineer developers: here's what enterprises should knowventurebeat.com
Anthropic’s AI used fake identities, malware in rogue attack on GitHub projectarstechnica.com
Yes, the AIs were given a cybersecurity challenge, with internet access enabled and safety filters d...
- Aug 7
- Aug 31
Anthropic by email
Get an email when Anthropic has news
No news that day, no email.
More stories today
- Developer open-sources LinkedIn prospect research tool as Claude Code plugin
- Polimill builds Japan's next-gen public AI infrastructure
- How Matic got robots into 10,000 homes
- Connect AgentCore MCP server to Amazon Quick
- Vercel's fx coding agent now supports Grok & Codex subs