Tools AI-Found Vulnerability Register aifv-0068
Register entry
Anthropic attributes 500 or more high-severity vulnerabilities that survived decades of scrutiny to Claude Opus
- Project
- open-source and partner codebases (unnamed)
- Finder
- Claude Opus (Anthropic)
- Autonomy
- model-assisted discovery per Anthropic's cybersecurity page
- Evidence tier
- Aggregate claim
- Bounty
- None published
- Confidence
- CONFIRMED
What happened
A separate figure from the Glasswing count, filed under Claude Opus rather than Mythos, with no identifier list.
Credit as recorded: "found that survived decades of scrutiny and automated analysis", filed under Claude Opus on Anthropic's cybersecurity page. Date basis: YFarmX report of 21 August 2026 quoting the page.
Sources
- YFarmX, Claude Security now scans on Mythos 5yfarmx.com/claude-security-scans-run-on-mythos-5/
One record from the AI-Found Vulnerability Register, maintained by the Security Desk. Data: CSV · JSON ·RSS · CC BY 4.0 with attribution to YFarmX.Tracker updated · 19 September 2026