Anthropic has launched Project Glasswing — a controlled initiative giving select organizations, including AWS, Apple, Cisco, Google, JPMorgan Chase and Microsoft, access to its unreleased frontier model Claude Mythos Preview to find and fix critical software vulnerabilities before malicious actors can exploit them.
The Programme
- Project Glasswing: controlled early access to Claude Mythos Preview
- Partners: AWS, Apple, Cisco, Google, JPMorgan Chase, Microsoft
- Goal: surface + patch critical vulnerabilities pre-exploit
Why It Matters
- Frontier AI turned toward defensive cybersecurity
- Deepens Anthropic’s enterprise + security positioning
- Signals confidence in Mythos’s code-reasoning ability
What To Watch
- Vulnerabilities found + disclosed responsibly
- Wider Claude Mythos release timing
- Rival labs’ security plays
Follow Vibes Uncut Media for continuing AI + security coverage.
Leave a Reply