More than 40 bitcoin and crypto firms asked the biggest AI labs this week to let independent security researchers use their strongest models before public release.
The request was laid out in a letter published on August 10. Its argument is built around a simple imbalance: the Bitcoin network secures more than $1 trillion, and the wider stack around wallets, signing devices, libraries, and custody systems holds trillions more. That code is developed in the open, and a single bug can wipe out someone’s savings.
The signatories say the maintainers keeping that infrastructure running cannot get into the vetted researcher programs operated by AI labs. When they turn to public models instead, the guardrails can shut down legitimate efforts to search for flaws. As a result, they rely on open-weight systems they can run themselves. The letter says those systems are useful, but still behind frontier models—and behind the hackers targeting cryptocurrency networks.
“It is clear that early access and reasonable compute budgets would help core developers produce more secure code and identify and remediate vulnerabilities before losses occur,” the letter reads. “We, therefore, ask frontier AI labs to establish or expand standing trusted-access programs for qualified defenders of open-source financial infrastructure.”
The letter also says labs and a small group of partners get a months-long preview of new attack capabilities before the rest of the world, while those capabilities still end up spreading through public releases and compromised systems.
AI labs have not made a public commitment
As of today, the major AI labs have not shared any public stance on this specific issue. The companies are, however, working on stronger safeguards intended to stop their models from being used for malicious purposes, whether in blockchain coding or elsewhere.
Late last month, the Bitcoin Security Consortium committed $15 million to the network’s long-term security. Its members include BlackRock, Coinbase, Strategy, Anchorage, ARK, Block, Blockstream, Fidelity Digital Assets, and Galaxy.
The new letter is aimed at a different gap. It is not asking for money to prepare for tomorrow’s threats, but for access to today’s models for the maintainers trying to patch them now.
Anthropic has shown how the same models can enable what it calls “vibe hacking,” describing criminals carrying out live bitcoin-ransom attacks with AI assistance, and has said the tools make offense cheaper. Giving stronger models to defenders would require labs to decide who qualifies and to trust those users not to leak capabilities.
The letter asks the labs to accept that risk anyway.
