More than three dozen Bitcoin and crypto companies are calling on leading AI labs to give open-source security researchers early access to their most powerful models. The companies say researchers defending crypto infrastructure are often forced to work with less capable tools than the attackers targeting the same systems.

The request was organized by the Bitcoin Policy Institute and published earlier this week. Coinbase, Block, BitGo, Blockstream, Anchorage Digital, ARK Invest, Bitwise, Foundry, Casa and Exodus are among the signatories. Developer-focused nonprofits including Brink, Chaincode and Btrust have also endorsed the letter.

The main concern centers on Bitcoin Core developers, the small group responsible for maintaining the software that supports the Bitcoin network. The companies argue that these developers generally cannot access the advanced AI models that labs make available to selected security partners.

Using public AI systems creates another challenge. Safety mechanisms intended to prevent malicious activity can also restrict legitimate cybersecurity research, including attempts to identify vulnerabilities before they are discovered and exploited by criminals.

Researchers are consequently pushed toward open-weight models that are freely available but generally less powerful than the latest proprietary systems.

According to the letter, attackers do not face the same restrictions. AI labs and a limited number of trusted partners can gain access to emerging offensive capabilities well before the public, while similar capabilities can eventually spread through publicly available models, compromised corporate accounts and specialized hacking tools.

Crypto Industry’s Five Requests

The signatories are seeking five key changes from AI labs. They want early access to advanced cybersecurity models, including systems that have not yet been publicly released; adequate computing resources for large-scale security reviews; protected environments for analyzing private code; programs that allow independent researchers and small development teams to participate; and direct communication channels with AI labs’ security teams for reporting vulnerabilities.

The appeal comes after recent incidents demonstrated how quickly AI is changing the cybersecurity landscape.

BTCPay Server, one of the signatories, recently disclosed a critical vulnerability that attackers had already exploited to drain merchants’ Lightning nodes. Foundation, a hardware wallet company that also signed the letter, lost one of its nodes during the attack.

BTCPay later highlighted AI’s growing impact on cybersecurity, noting that advanced models can make it faster and cheaper to examine large codebases for potential weaknesses.

In this instance, the vulnerability was uncovered by defenders rather than attackers. A volunteer organization known as the Bitcoin Red Team began using AI models to analyze Bitcoin-related codebases earlier this month and has submitted thousands of findings covering hundreds of projects.

One of those investigations resulted in the vulnerability report that led to BTCPay Server’s subsequent security patch.