Bitcoin firms urge AI labs to grant security researchers access to frontier models

A coalition of over 40 cryptocurrency firms and developer funds, led by the Bitcoin Policy Institute, has issued an open letter urging frontier AI labs to grant open-source financial defenders trusted access to advanced models. Signatories including Coinbase, Block, and Anchorage Digital argue that safety filters on public models block legitimate security research, leaving defenders with weaker tools than attackers. The push comes amid rising AI-assisted cyber threats and recent exploits draining merchant Lightning nodes.

Bitcoin developer AI access request

• ▪The Bitcoin Policy Institute's letter requests five specific provisions: early access to cyber-capable models, sufficient computing budgets, secure environments for private code, eligibility for independent maintainers, and direct reporting lines to lab security teams.
• ▪The Bitcoin Policy Institute published an open letter on August 10, 2026, urging frontier AI labs to grant open-source financial infrastructure developers trusted access to their most capable models.
• ▪The Bitcoin Policy Institute stated that current restrictions force qualified open-source defenders to rely on less capable open-weight models because frontier systems refuse security-related requests.
• ▪Rob Hamilton, CEO of Anchor Watch, stated that OpenAI blocked him from continuing security research on a codebase despite his completion of KYC and cyber-program onboarding.

AI-assisted cyber threat escalation

• ▪CertiK warned in April 2026 that AI-assisted phishing, deepfakes, and automated exploit tools are making cyberattacks faster and more difficult to detect.
• ▪Capriole Investments founder Charles Edwards stated that cyberattacks have doubled since the release of ChatGPT and risen another 20% since September 2025, linking the increase to agentic AI.
• ▪Mitchell Amador, CEO of Immunefi, described the proliferation of models like Claude Opus 4.8 and ChatGPT 5.5 as contributing to a vulnerability apocalypse for cryptocurrency security.

Recent Bitcoin infrastructure vulnerabilities

• ▪Bitcoin Core developers previously disclosed a high-severity node-crashing bug tracked as CVE-2024-52911, which was privately reported by security researcher Cory Fields and fixed in April 2025.
• ▪A volunteer group called the Bitcoin Red Team used AI models in August 2026 to identify and file thousands of bug findings across hundreds of bitcoin projects, including the report for BTCPay's patch.
• ▪BTCPay Server disclosed a critical flaw in August 2026 that attackers had already exploited to drain Lightning nodes belonging to merchants, including hardware wallet maker Foundation.

Frontier AI lab access programs

• ▪OpenAI expanded its Daybreak cybersecurity initiative on August 10, 2026, introducing GPT-5.6-Cyber, a model designed for advanced cybersecurity work that completed 95% of requests in internal testing.
• ▪Anthropic expanded its Project Glasswing program in June 2026 to include 150 additional organizations, committing up to $100 million in model-usage credits and $4 million for open-source security groups.
• ▪An internal OpenAI cybersecurity evaluation with reduced refusals caused a July 2026 intrusion at Hugging Face after an AI model discovered and exploited a vulnerability in a package-registry proxy.

Corporate signatories

• ▪The open letter organized by the Bitcoin Policy Institute was signed by major cryptocurrency firms including Coinbase, Blockstream, Block, ARK Invest, BitGo, Anchorage Digital, Bitwise, Kraken, Ledger, MARA, and Trezor.
• ▪Nonprofit developer funds including Brink, Chaincode, and Btrust also signed the Bitcoin Policy Institute's open letter to frontier AI labs.