Over 40 bitcoin and crypto companies request access to AI labs' frontier models for independent security research before public release.
AI & Agents ·
More than 40 bitcoin and crypto firms, including Coinbase, Block, and BitGo, requested that major AI laboratories establish trusted-access programs allowing independent security researchers to use frontier models before public release. The Bitcoin Policy Institute organized the effort, with signatories spanning Blockstream and ARK Invest among others. The request targets a practical vulnerability: developers maintaining open-source cryptocurrency infrastructure—wallets, node software, and custody systems—cannot access the most advanced AI models, while attackers already can.
The letter, published August 10, highlights the asymmetry created by security guardrails on public models that block legitimate vulnerability hunting. Developers currently rely on open-weight systems they can run locally, which remain a step behind frontier capabilities. The signatories argue that early access and reasonable compute budgets would enable core developers to produce more secure code and identify vulnerabilities before losses occur, particularly given that the Bitcoin network and wider stack secure trillions of dollars.
As of the letter's publication, the major AI laboratories have not made public commitments on the proposal. The effort follows a separate $15 million security commitment announced last month by the Bitcoin Security Consortium, which includes BlackRock, Fidelity Digital Assets, and Galaxy. Whether frontier labs will establish or expand such programs remains unresolved.