How many vulnerabilities did Anthropic's AI find in Firefox?

Anthropic's AI identified a total of 22 vulnerabilities in Firefox.

Were the high-severity bugs in Firefox fixed?

Yes, most of the 14 high-severity bugs found by AI have been fixed in Firefox 148.

Did AI help in creating exploits for Firefox vulnerabilities?

The AI was effective in finding vulnerabilities but struggled to create exploit proofs-of-concept, with success in only two cases.

Home / Technology / AI Discovers Firefox Flaws: 14 High-Severity Bugs Found

AI Discovers Firefox Flaws: 14 High-Severity Bugs Found

7 Mar

•

Summary

AI identified 22 vulnerabilities in Firefox, with 14 being high-severity.
Most bugs are fixed in Firefox 148, released recently.
AI excelled at finding flaws but struggled to create exploit proofs-of-concept.

AI Discovers Firefox Flaws: 14 High-Severity Bugs Found

In a significant security audit, Anthropic's AI tool, Claude Opus, identified 22 vulnerabilities within the widely-used Firefox browser. The analysis, which spanned two weeks and began within the javascript engine, uncovered 14 of these flaws as high-severity.

These issues have largely been rectified in Firefox 148, a release that became available this February. However, a small number of fixes are slated for the subsequent software update, indicating the ongoing nature of security patching.

The AI's capability shone in vulnerability detection rather than exploit creation. The Anthropic team reportedly spent $4,000 on API credits attempting to build exploit demonstrations, with limited success in only two instances. This highlights the dual nature of AI in open-source projects, offering powerful discovery tools alongside potential challenges.

Disclaimer: This story has been auto-aggregated and auto-summarised by a computer program. This story has not been edited or created by the Feedzop team.