genai / news / / The Economic Times
Mythos Preview was used to scan more than 1,000 open-source software projects.
Anthropic's Mythos Preview AI found over 10,000 high- or critical-severity vulnerabilities in open-source projects.
KEY POINTS
- Security firms confirmed 90.6% validity of Mythos Preview’s 1,752 flagged vulnerabilities, with 62.4% as high or critical risk.
- Cloudflare discovered 2,000 bugs in critical-path systems with Mythos, reporting fewer false positives than human testers.
- Mozilla fixed 271 Firefox vulnerabilities using Mythos, which outperformed Anthropic’s earlier Claude Opus 4.6 model.
- Anthropic warns that human teams may be unable to keep pace with AI-discovered vulnerabilities.
COMPANIES
Summarized by Newsio from The Economic Times. How we summarize →