Bitcoin Red Team flags dozens of critical flaws as AI-driven security campaign scales up

The volunteer Bitcoin Red Team says it has scanned 150 repositories and privately disclosed more than a dozen vulnerabilities, reporting a rapid rise in critical findings as it works to harden wallets, cryptographic libraries and broader infrastructure in the wake of the COLDCARD exploit. The group estimates the effort has already used about $20,000 in AI compute, with OpenSats covering the bill. Researchers said they are running multiple AI toolchains, and many now view open-source models such as Kimi K3 as the main workhorse for security research. By contrast, they described frontier closed models from firms including OpenAI and Anthropic as more constrained for this kind of analysis. The team plans to open-source its AI security harness to help projects continuously audit their own codebases. Researchers also warned Bitcoin may be the first sector to see a surge in AI-assisted vulnerability discovery, a pattern they expect to spread across open-source software more broadly.