Overview
- The Bitcoin Policy Institute led a coalition of more than 40 digital-asset firms that published an open letter on Aug. 10 asking major AI laboratories to give qualified open-source security researchers trusted, pre-release access to frontier models.
- The coalition requested specific resources: early model access, sufficient compute for long-running reviews, secure research environments to keep sensitive code private, and direct channels to AI lab security teams for coordination.
- At the time of reporting, neither OpenAI nor other leading labs had publicly implemented a program that exactly matches the coalition’s controlled-access proposal, though some labs have separate limited cyber-access initiatives.
- The appeal follows concrete failures and misuse: a Coldcard firmware build error reportedly weakened seed entropy and led to large thefts, researchers said North Korea-linked Kimsuky runs local models for malware and phishing, and volunteer AI reviews flagged thousands of potential Bitcoin-related issues.
- Industry funding for Bitcoin security has grown, but continuing losses and attackers’ use of open or local models mean the coalition’s push could change how labs balance safety controls with practical needs for forensic and defensive work.