Anthropic AI finds major Firefox security flaws

A new artificial intelligence (AI) system developed by Anthropic has reportedly uncovered a large number of security vulnerabilities in an unreleased version of the Firefox browser, raising fresh discussions about the future of cybersecurity and automated code analysis.

According to reports, the company's restricted "Mythos Preview" model was used in collaboration with Mozilla to examine the source code of the upcoming Firefox version 150. The system identified around 271 potential security flaws before the software was publicly released.Pentagon Adopts Palantir AI System After Dispute With AnthropicPentagon Adopts Palantir AI System After Dispute With Anthropic

Mozilla stated that the results highlight a significant leap in artificial intelligence-assisted security testing. The company's Chief Technology Officer, Bobby Holley, said the findings demonstrate how advanced AI tools can give defenders an advantage over attackers by identifying complex vulnerabilities much faster than traditional methods.

He compared the performance with earlier AI tools, noting that a previous model had detected only a small fraction of issues in an earlier Firefox version. The new system, however, was able to analyze far deeper layers of the codebase and uncover hidden risks that might have taken human researchers months to detect.

The development has intensified debate within the technology industry over the role of powerful AI systems in cybersecurity. Some experts argue that such tools could revolutionize digital defense by making software significantly more secure, especially in widely used open-source projects.

Others caution that the same capabilities could potentially be misused if placed in the wrong hands, increasing concerns about automated hacking or large-scale vulnerability discovery by malicious actors.

Mozilla executives emphasized that modern software development is becoming increasingly dependent on advanced AI-based auditing tools. They also called for broader access to such systems for trusted developers working on critical open-source infrastructure.

As AI continues to evolve, experts say the balance between security enhancement and potential risk will remain a central challenge for the tech industry.