Anthropic’s Claude AI discovered 22 Firefox vulnerabilities, including 14 high-severity flaws, prompting rapid security fixes.
Artificial intelligence is increasingly playing a role in cybersecurity research after an AI model helped uncover dozens of previously unknown security flaws in the Firefox web browser.
Researchers at Anthropic revealed that their Claude Opus 4.6 AI model discovered 22 vulnerabilities in Firefox during a two-week testing period as part of a security collaboration with Mozilla. Of those issues, 14 were classified as high-severity, while the rest were rated moderate or low risk.
The vulnerabilities were identified by allowing the AI system to analyze Firefox’s source code. During the project, the model scanned nearly 6,000 C++ files and generated more than 100 bug reports for Mozilla’s engineering team.
One of the first findings occurred just minutes into the testing process, when the system detected a “use-after-free” memory issue in the browser’s JavaScript engine. Human researchers later verified the discovery to confirm it was a legitimate security flaw rather than a false positive.
Mozilla worked with Anthropic to review the findings and implement fixes. Most of the vulnerabilities were patched in Firefox version 148, released earlier this year, while additional fixes are expected in upcoming updates.
Despite its ability to identify vulnerabilities, the AI system showed limited success when attempting to transform the bugs into real-world exploits. After hundreds of attempts, the model produced only two proof-of-concept exploits, neither of which could bypass Firefox’s real-world security protections.
Security researchers say the experiment demonstrates how artificial intelligence could dramatically accelerate vulnerability discovery. However, it also raises concerns that similar tools could eventually be used by malicious actors to find software weaknesses faster than developers can fix them.
Reports are sourced from official documents, law-enforcement updates, and credible investigations.
Discover additional reports, market trends, crime analysis and Harm Reduction articles on DarkDotWeb to stay informed about the latest dark web operations.
