Crypto Firms Urge AI Developers to Share Powerful Models with Security Researchers

Crypto firms are calling on AI developers to provide security researchers with early access to their most powerful models, citing rising hacking threats.

The open letter, organized by the Bitcoin Policy Institute, was signed by over 30 companies including Coinbase, Block, BitGo, and Anchorage Digital. The signatories argue that existing restrictions make it harder for defenders to identify vulnerabilities while attackers can bypass these barriers.

Researchers currently face safety restrictions when using publicly available systems, which can block legitimate attempts to identify vulnerabilities. This forces security specialists to rely on open-weight models, which may be less capable than the most advanced closed systems.

The signatories proposed giving researchers early access to powerful AI models, sufficient computing resources, and secure environments for analyzing private code. They also want such programs to be available to smaller teams and direct communication channels established with AI companies' security teams.