Palo Alto Networks internal testing showed frontier AI models, accessed early, matched one year of manual penetration testing in three weeks. The company stated these models identify vulnerabilities across vast, complex codebases with high effectiveness.
The findings indicate AI is transitioning from a helpful assistant to an autonomous agent. This technology can discover and chain multiple security flaws at a scale that may challenge existing defense strategies.
This capability will accelerate vulnerability discovery for both attackers and defenders. Consequently, the time window for patching critical flaws will shrink significantly.