Artificial intelligence, through the Claude Opus model, has identified 22 vulnerabilities in the Firefox code.

Artificial intelligence is making a simulated presence in cybersecurity research, exemplified by the collaboration between Anthropic and Mozilla. The AI model Claude Opus discovered 22 vulnerabilities in the Mozilla Firefox code during a two-week testing project, of which 14 were classified as having high severity. Most of these issues have already been fixed in Firefox version 148, released this year, and a few solutions are expected in future updates. The project highlights how AI systems can assist security teams in reviewing complex software environments.

Researchers at Anthropic began the analysis with Firefox's JavaScript engine, later expanding to other parts of the code. Firefox was chosen due to its status as a well-tested open-source project, ideal for assessing the effectiveness of vulnerability discovery using AI. Although Claude was effective in identifying security flaws, it faced difficulties in developing functional exploits, spending approximately $4,000 in API credits to generate proof of concept, succeeding in only two cases. Nevertheless, the experiment underscores the growing potential of AI as a cybersecurity tool, especially in open-source ecosystems, where AI-based analysis could significantly accelerate vulnerability discovery.