Crypto news

The Bitcoin Policy Institute, together with a broad coalition of digital asset companies, has published an open letter to the world's largest AI labs. The authors demand that verified open-source developers be granted reliable access to cutting-edge artificial intelligence models to protect financial infrastructure.

The letter has been signed by dozens of organizations from the digital asset, cybersecurity, and open-source sectors. These include developers, major custodians, payment services, and security firms whose business depends on the integrity of open infrastructure and libraries.

Infrastructure defenders are left without the strongest tools

In the open letter, the authors note that the need for such access has become evident over the past few weeks. People who protect digital asset infrastructure and open-source software need the latest AI capabilities to conduct thorough security audits. Without them, it is difficult to stay ahead of increasingly sophisticated attackers.

Open source has long become financial infrastructure. According to the authors' estimates, the Bitcoin network alone safeguards more than a trillion dollars, and across the entire digital asset ecosystem—wallets, signing devices, cryptographic libraries, node software, custodial systems, exchanges, and payment networks—trillions of dollars depend on it. This infrastructure is openly developed by companies, non-profit organizations, independent researchers, and grant-funded maintainers.

Behind this software are the personal funds of millions of people. These are pension savings, business capital, reserve funds, and family wealth, so a single vulnerability can wipe out a lifetime of savings. Digital assets attract cyberattacks precisely for this reason.

Many cryptocurrencies are structured as bearer assets, so the path from a hack to lost funds is short. In such conditions, it is especially important to find and close vulnerabilities in advance, the letter's authors emphasize. Cutting-edge AI is meanwhile changing the economics for both security researchers and attackers.

Defenders occupy the least privileged position

Advanced models are capable of reviewing large codebases, identifying potential weaknesses, and accelerating complex technical work—for both defenders and attackers alike. The authors expect these capabilities to continue spreading through public open-source models, unauthorized access to corporate systems, and specialized cyberattack tools.

Today, leading labs and select partners see new cyber capabilities months before they reach the rest of the ecosystem. At the same time, access pathways to such tools for qualified defenders remain limited and inconsistent. Many digital asset developers, including Bitcoin Core maintainers, are excluded from lab programs.

Restrictions on public frontier systems often block legitimate security work. As a result, defenders are forced to conduct critical audits on less powerful open-weight models. According to the letter's authors, developers occupy the most vulnerable position in the entire security ecosystem around AI.

The Bitcoin Policy Institute received several independent reports from open-source maintainers. According to them, experienced attackers, possibly from other countries, are conducting continuous AI-assisted attacks, and small teams find it difficult to sustain such a pace.

Even when systems remain protected, responding to persistent and sophisticated activity diverts the maintainers' scarce resources away from development and routine security work. Bitcoin developers and other security researchers are working around the clock to audit wallets, libraries, protocols, and open tools using available open-weight models, the authors add.

What the coalition is asking of the labs

Early access and reasonable compute budgets would help key developers write more secure code and find and fix vulnerabilities before losses occur. The coalition is asking frontier AI labs to create or expand standing trusted access programs for qualified defenders of open financial infrastructure.

Such programs should provide early, controlled access to the strongest cyber models and, where appropriate, pre-release versions. Among the conditions, the authors also cite "sufficient" computing power and the ability of autonomous agents to work on long-running tasks—without which meaningful security audits cannot be conducted.

Separately, the coalition asks for secure environments for analyzing private or closed-source code, as well as participation rights for small, non-profit, and independent maintainers on par with large companies and states. The final item on the list is a direct channel for coordinated vulnerability reporting, remediation, and contact with lab security teams.

The digital asset industry is ready to help identify trusted participants, establish standards, and coordinate remediation when issues are found. Frontier AI, in the authors' view, could become one of the most powerful defensive technologies in history—but only with fair access to such systems.

The letter was signed by representatives of Bitcoin Policy Institute, Coinbase, Block, Blockstream, BitGo, Anchorage Digital, Bitwise, Trezor, MARA, Foundry, Galaxy, Strategy, Strike, and River, as well as dozens of other participants from the digital asset, cybersecurity, and open-source sectors.

My take: this step is a logical continuation of AI's evolution as a dual-use tool. While labs grant access to select corporate partners, defenders of open infrastructure remain in a knowingly losing position. If the industry truly wants to prevent catastrophic hacks, it will have to acknowledge that the security of decentralized systems is a shared responsibility, and access to frontier models must be symmetric—not a privilege of large players.