Mozilla paid $4,000 to find bugs in Firefox using AI
Technology
Anthropic's new AI, Claude Opus 4.6, teamed up with Mozilla and managed to uncover 22 security vulnerabilities in Firefox, all in just two weeks.
This experiment was all about testing if AI can help find tricky bugs that humans might miss.
Mozilla has started using Claude internally for security purposes
Out of the 22 issues, 14 were classified as high-severity; Anthropic said most of the reported issues were fixed in Firefox 148.0, with the remainder slated for upcoming releases.
The project cost $4,000 in API credits, and Mozilla has started using Claude internally for security purposes.