A coalition of more than three dozen cryptocurrency and bitcoin firms is sounding the alarm over a growing technological gap between digital asset defenders and the hackers targeting them. In a collective appeal organized by the Bitcoin Policy Institute, industry heavyweights including Coinbase, Block, and ARK Invest are urging major artificial intelligence laboratories to grant open source security researchers early access to their most sophisticated models. Their core argument is simple but urgent: those tasked with protecting trillions of dollars in global infrastructure are currently fighting with inferior tools compared to the adversaries they are trying to stop.
The frustration stems from a restrictive environment where Bitcoin Core developers often find themselves locked out of exclusive trusted partner programs managed by AI labs. When these engineers attempt to use public versions of high powered AI, they frequently hit safety filters intended to prevent the creation of malware. While these guardrails are meant to protect the general public, they inadvertently hinder legitimate security experts who need to simulate attacks to identify and patch vulnerabilities before criminals can exploit them. This forces many defenders onto smaller, open weight models that lack the reasoning capabilities of the top tier proprietary systems.
This imbalance was recently highlighted by a series of exploits involving BTCPay Server and various Lightning nodes, which resulted in drained assets for several merchants_ These incidents serve as a stark reminder that attackers operate without ethical constraints or filtered interfaces. However, there is a silver lining; a volunteer group known as the Bitcoin Red Team successfully identified critical flaws by deploying AI against bitcoin codebases, filing thousands of reports across hundreds of projects. This proves that while AI makes searching for weaknesses faster and cheaper for everyone, the advantage currently tilts toward those with unrestricted access.
To level the playing field, the signatories are requesting specific concessions from AI developers, including pre release access to cyber capable models and dedicated computing budgets for thorough reviews. They are also pushing for secure environments where private code can be examined without risk and direct communication lines into lab security teams. Crucially, they want these opportunities opened up to small independent maintainers rather than just massive corporations, ensuring that every layer of the network’s defense is reinforced regardless of company size.
