Security researchers warn that AI models are making Bitcoin software increasingly vulnerable to automated exploitation.
Tech & Launches ·
A group of roughly 20 to 25 developers calling itself the Bitcoin Red Team has begun systematically scanning Bitcoin's open-source software layer to identify vulnerabilities before AI-powered attackers can exploit them. The team formed in response to what member Calle described as an urgent threat: increasingly capable and accessible artificial intelligence models now allow people with minimal security expertise to discover and execute exploits. The effort was spurred partly by the Coldcard hardware wallet breach and the availability of more powerful Chinese AI models, which Calle said have given both defenders and malicious actors unprecedented capabilities.
The Red Team works both reactively—responding to projects seeking security audits—and proactively, having already surveyed nearly all significant Bitcoin ecosystem repositories without being asked. Calle emphasized that while Bitcoin's core protocol remains secure, the ecosystem of applications, wallets, and services built atop it presents the real risk surface. The team has spent approximately $20,000 across various security-scanning services and shares findings directly with developers to refine vulnerability classifications.
Chinese AI models dominate the Red Team's work because American models typically block cybersecurity requests, leaving a gap that more permissive systems fill. What remains unclear is whether the Red Team's proactive scanning can maintain pace with the acceleration of AI-assisted threat discovery, or whether attackers are already ahead.