Coinbase, Block, BitGo, and 70+ other crypto companies say frontier AI safety guardrails hinder legitimate security work while cyberattackers use stronger tools
More than three dozen bitcoin and crypto companies have asked the largest AI labs to give open-source security researchers early access …
Context & Ripple Effects
The request brings crypto companies into an access-governance dispute that has been building since researchers sought a legal and technical safe harbor to study AI systems. It frames security research as a distinct use case rather than a general argument for unrestricted model release.
The timing also extends a security-focused coalition agenda: Chainguard, Cisco, Cloudflare, JPMorgan Chase and others recently formed Athena to use AI in securing open-source software. Meanwhile, companies defending open-weight models have argued that restrictions can weaken competition and cybersecurity.
First-order effects
- Coinbase, Block, BitGo and their peers have put AI labs’ researcher-access policies under direct pressure, asking for earlier access for open-source security work.
- The companies’ complaint makes the asserted gap between constrained defenders and better-equipped cyberattackers central to the labs’ safety-guardrail debate.
Second-order effects
- Athena’s open-source security agenda gains another industry constituency, strengthening the case that vetted security research needs treatment separate from broad public model access.
- The request sharpens the practical divide in the open-weight debate: firms advocating fewer restrictions can point to crypto security work, while frontier labs must defend why their controls apply to that use case.
Third-order effects
- If AI labs create researcher-specific pathways, model access will increasingly function as a security boundary managed by user category and purpose, rather than a simple choice between public release and refusal.
- The recurring safe-harbor, open-weight and crypto-security arguments point toward frontier-model governance being negotiated through sector-specific security claims as well as general safety policy.
The trend: Frontier-model access governance is moving toward pressure for differentiated access for vetted security researchers, rather than uniform guardrails for every user.