Key facts
- Crypto companies and the Bitcoin Policy Institute are asking AI firms for early access to advanced AI models.
- The request aims to bolster cybersecurity efforts for Bitcoin and other open-source financial infrastructure.
- Signatories argue that current public AI models have limitations for security research.
- The letter highlights that vulnerabilities in open-source infrastructure can endanger life savings.
- Reports suggest advanced AI is already being used by sophisticated actors to target open-source maintainers.
A coalition of cryptocurrency companies, led by the Bitcoin Policy Institute (BPI), has formally requested that leading artificial intelligence (AI) firms grant early access to their most advanced models. The appeal, detailed in a letter published on Monday, aims to equip developers of critical open-source financial infrastructure, such as Bitcoin, with superior cybersecurity tools.
The BPI highlighted that many defenders of digital assets, including Bitcoin Core developers, are currently restricted from accessing proprietary cyber programs and face limitations with publicly available frontier AI systems. This forces them to rely on less capable open-weight models, potentially leaving vital infrastructure vulnerable.
The signatories, which include prominent names like Kraken, Ledger, Trezor, and BitGo, urged AI labs to establish or expand 'trusted-access programs' for qualified defenders of open-source financial infrastructure. They emphasized that open-source software underpins significant digital and financial systems, with Bitcoin alone securing over $1 trillion in value, and that vulnerabilities can put individuals' life savings at risk.
The letter argues that frontier AI is fundamentally altering the landscape of security research and cyber operations. These advanced models possess the capability to analyze vast codebases, identify potential weaknesses, and accelerate complex technical tasks for both offensive and defensive actors. The BPI suggested that frontier AI could become one of the most powerful defensive technologies ever developed, but stressed that without dedicated access programs, defenders may struggle to keep pace with evolving threats.
Concerns are amplified by reports from open-source maintainers who have described sophisticated entities, potentially including foreign adversaries, leveraging advanced AI to sustain attacks. The rise of AI-assisted vulnerability discovery has led to significant apprehension within the crypto security industry, with new models like Claude Opus 4.8 and ChatGPT 5.5 contributing to what Immunefi CEO Mitchell Amador termed a 'vulnerability apocalypse' for the sector.