Claude's Firefox Audit Redefines What "Comprehensive" Security Testing Means

Anthropic's Claude discovered 271 vulnerabilities in Firefox during a two-week audit, including 14 high-severity flaws representing one-fifth of all high-severity bugs Mozilla patched in 2025. Analysis: The speed and scale suggest that traditional human-led audits—even on extensively reviewed, open-source codebases—may no longer constitute comprehensive testing. Organizations now face an uncomfortable question: if mature software still harbors hundreds of undiscovered flaws after years of scrutiny, what gaps exist in their own security practices.