Anthropic’s Claude Opus 4.6 identifies critical vulnerabilities, highlighting the growing role of AI in cybersecurity
San Francisco, United States, 10 March 2026 – Artificial intelligence is becoming an important tool in cybersecurity, helping researchers identify software vulnerabilities faster than ever before. In a recent experiment, AI company Anthropic revealed that its latest large language model, Claude Opus 4.6, discovered multiple security vulnerabilities in Mozilla’s Firefox browser within a short period of time.
According to Anthropic, the company collaborated with Mozilla’s security researchers to analyze the browser’s source code. Over two weeks, the AI model scanned nearly 6,000 C++ files from Firefox’s codebase to detect potential security weaknesses.
During the experiment, Claude Opus 4.6 identified 22 vulnerabilities in the Firefox browser. Mozilla later classified 14 of these as high-severity issues that could potentially pose serious security risks if left unaddressed.
In total, Anthropic submitted 112 unique vulnerability reports to Mozilla. Many of these issues have already been fixed in Firefox version 148, while the remaining vulnerabilities are expected to be addressed in upcoming software updates.
Anthropic described the findings as a significant step forward in AI-powered cybersecurity. The company noted that in just two weeks, its AI system was able to uncover nearly one-fifth of the high-severity vulnerabilities that Mozilla resolved throughout the entire year of 2025.
Claude Opus 4.6 is the newest version of Anthropic’s advanced large language model, launched in February 2026. The company has positioned the tool as a valuable asset for cybersecurity professionals, claiming that it is especially effective at identifying critical software vulnerabilities.
One of the key advantages highlighted by Anthropic is the AI model’s ability to analyze large and complex codebases without requiring specialized prompts or customized tools. In this case, the system was able to detect issues directly while reviewing open-source software files.
Firefox was chosen for the experiment because it is widely considered one of the most secure and thoroughly tested open-source browsers available. By targeting a complex and well-maintained project, the researchers aimed to demonstrate how AI can support even the most mature cybersecurity environments.
The experiment also highlights a broader trend in the technology industry. As software systems grow larger and more complex, companies are increasingly using artificial intelligence to strengthen cybersecurity practices and speed up vulnerability detection.
AI-driven security tools can help developers identify risks earlier in the development process, reduce manual code reviews, and improve overall software reliability. Many cybersecurity experts believe that combining human expertise with AI analysis will become a standard approach to protecting modern digital infrastructure.
As artificial intelligence continues to evolve, its role in software security, threat detection, and vulnerability management is expected to grow rapidly.