A coalition of digital asset organizations, led by the Bitcoin Policy Institute, is calling on frontier AI laboratories to provide qualified open-source security defenders with trusted access to advanced AI models. The group argues that developers need more powerful tools to audit financial software, wallets, cryptographic libraries, and other critical infrastructure across the cryptocurrency ecosystem.

The coalition is seeking secure testing environments, sufficient computing capacity, and direct communication channels for coordinated vulnerability disclosure. According to the Bitcoin Policy Institute, open-source developers frequently lack access to advanced cybersecurity models, and public frontier systems can impede legitimate security research through their built-in safety controls. Frontier labs—companies such as OpenAI, Anthropic, and Google DeepMind that develop the most capable AI systems—have implemented safety guardrails that can refuse or restrict queries involving code analysis, exploit identification, and vulnerability testing, even when the request comes from a legitimate defender.

Security Access Gap Identified

As a consequence of these restrictions, some defenders are forced to rely on less capable open-weight models for security reviews. The coalition contends this creates an asymmetry between defenders and malicious actors who may already be using stronger AI capabilities.

The group emphasized that the issue affects infrastructure underpinning Bitcoin and the broader digital asset sector. That infrastructure includes signing devices, cryptographic libraries, custody systems, exchanges, and payment networks. The institute noted that Bitcoin alone secures more than $1 trillion in value, while the wider digital asset ecosystem encompasses trillions of dollars across related infrastructure. Much of this infrastructure is maintained by small open-source teams or volunteer contributors, a structure that contrasts with enterprise software development where dedicated security budgets and dedicated security teams are standard.

AI as a Vulnerability Detection Tool

The coalition stated that advanced AI can scan large codebases and flag potential weaknesses, potentially accelerating complex security work for both defenders and attackers. However, frontier labs and selected partners currently gain access to emerging cyber capabilities before the broader user base, leaving qualified open-source defenders without comparable tools.

The Bitcoin Policy Institute reported that maintainers have encountered sophisticated attacks that place significant strain on small development teams. Some of these incidents, the institute said, involved potential foreign adversaries leveraging advanced AI capabilities.

Despite these limitations, developers have continued using available open-weight models to review wallets, protocols, and open-source tools. The institute said these efforts have already yielded actionable security findings and fixes.

Proposed Trusted Access Programs

The coalition is asking for earlier access to stronger models before attackers widely obtain similar capabilities, along with reasonable computing budgets to support long-running security reviews.

Under the proposed framework, programs would offer controlled access to advanced cyber-capable models, with pre-release access granted where appropriate and secure environments provided for reviewing private code. The coalition envisions eligibility extending to nonprofit organizations, independent maintainers, commercial companies, and government entities.

Direct communication channels with AI lab security teams would facilitate responsible disclosure and remediation. The digital asset industry indicated it can assist in identifying credible participants, establishing participation standards, and coordinating remediation efforts when security issues are discovered.