Anthropic demonstrates Claude discovering and exploiting real-world software vulnerabilities

Anthropic research published in October 2026 reported that an internal research model version of Claude identified and built exploit chains for five critical vulnerabilities in widely used software, including Chrome, affecting billions of users. Anthropic reported each vulnerability to vendors under coordinated disclosure, giving them 90 days to patch before public discussion of details beyond existing advisories. The demonstrations included remote code execution exploits requiring little to no user interaction.

Anthropic demonstrates Claude discovering and exploiting real-world software vulnerabilities According to Anthropic research published October 2026, an internal research model version of Claude identified and created exploit chains for five critical vulnerabilities in widely used software including Chrome, affecting billions of users. Anthropic reported each vulnerability to vendors under coordinated disclosure, providing 90 days to patch before public discussion of details beyond existing advisories. The demonstrations include remote code execution exploits requiring little to no user interaction. Topics Sources - Official Read article https://www.anthropic.com/research/discovering-vulnerabilities-with-claude Go deeper This intelligence is sourced automatically from public sources across the web and synthesised by the Prefactor AI pipeline. Stories are reviewed before publication.