Bitcoin Red Team Uncovers Dozens of Critical Vulnerabilities with AI

A volunteer initiative called Bitcoin Red Team has been using AI to scan over 150 Bitcoin repositories and uncover vulnerabilities. The team has spent around $20,000 on AI services and is developing an open-source AI platform for auditing Bitcoin software.

According to AnchorWatch CEO Rob Hamilton, the group uses various AI models including Kimi K3, OpenAI's GPT Sol, Anthropic's Claude Fable and Opus, and Z.ai's GLM 5.2 to identify vulnerabilities and generate supporting documentation.

The team has already reported more than a dozen vulnerability disclosures and is averaging one critical exploit per hour per person. The initiative has connected with OpenAI for help in running the Cyber Harness, a more expensive scan that has yielded good results.

Pseudonymous Bitcoin developer Calle said the effort is uncovering critical vulnerabilities across wallets, cryptographic libraries, and infrastructure. The team has burned through $10,000 per day and has secured funding, making further donations unnecessary.