Crypto Firms Ask AI Labs for Vetted Access to Frontier Security Models

More than three dozen crypto organisations have signed an open letter asking artificial intelligence laboratories to give vetted security researchers controlled access to their most capable cybersecurity models, according to CoinDesk's reporting.

The named signatories include Coinbase, Block, BitGo, Blockstream, Anchorage Digital, ARK Invest, Bitwise, Foundry, Casa and Exodus, alongside the nonprofit development organisations Brink, Chaincode and Btrust, plus BTCPay Server. The Bitcoin Policy Institute announced the coalition in an August 10 post on X, per crypto.news.

The letter sets out five asks: early access to the strongest cyber-capable models, including before public release; adequate compute budgets for security reviews; secure environments for examining private code; eligibility rules that reach small and independent maintainers rather than large firms alone; and direct channels to report findings to lab security teams.

The argument the signatories make is one of asymmetry. Public guardrails on advanced models, they say, block legitimate defensive research while attackers face no such constraint, so the people maintaining open-source infrastructure are working with weaker tools than the people attacking it. BTCPay Server's volunteer red team has filed thousands of findings across hundreds of projects, per CoinDesk.

What the letter does not ask for is worth stating plainly, because it is the part most easily misread: it does not seek the removal of safety filters, and it does not seek unrestricted public access to advanced cybersecurity capabilities. It asks for a vetted programme with oversight.

No AI laboratory has been reported as responding. This desk will carry any response when one is made rather than infer one from silence.