Anthropic’s Claude Opus 4.6 has reportedly identified 500 zero-day vulnerabilities, a claim that, while met with some skepticism, is being taken seriously by cybersecurity researchers. This development underscores the increasing potential of advanced AI models to contribute significantly to vulnerability detection and cybersecurity efforts.
Source: Simon Willison