Anthropic Finds 22 Firefox Vulnerabilities Using Claude Opus 4.6 AI Model
Summary: Anthropic said its Claude Opus 4.6 model found 22 new Firefox vulnerabilities over a two-week period in January 2026 - 14 high, seven moderate, and one low - which were addressed in Firefox 148. The company scanned nearly 6,000 C++ files, submitted 112 unique reports, and said the high-severity bugs it found amount to almost a fifth of all high-severity Firefox vulnerabilities patched in 2025.
Why it matters: Anthropic spent about $4,000 in API credits trying several hundred times to turn known Firefox vulnerabilities into working exploits and succeeded only twice - but it did succeed, including an exploit for CVE-2026-2796, a JIT miscompilation in JavaScript WebAssembly with a CVSS score of 9.8.
Read the original at The Hacker News ↗
- Topics
- AI Security
- Mentions
- CVE-2026-2796
- Related
- ⚡ Weekly Recap: Qualcomm 0-Day, iOS Exploit Chains, Air Snitch Attack & Vibe-Coded Malware Cyber Risks ·SECURITY AFFAIRS AI-CYBERSECURITY NEWSLETTER ROUND 8 Cyber Risks · Nov 15, 2026Pentagon Renames CDAO ‘Super Intelligence Office’ Cyber Risks · Nov 15, 2026