More than 40 Bitcoin and digital-asset organizations have urged leading artificial intelligence (AI) labs to grant vetted open-source security researchers controlled early access to advanced AI systems, warning that attackers currently hold the advantage. The appeal appears in an open letter titled “Defenders Need the Frontier,” published by the Bitcoin Policy Institute on Aug. 10 and signed by firms including Coinbase.
Open Letter Calls for Early Access for Defenders
The letter argues that as cutting-edge AI models grow more capable, malicious actors can rapidly weaponize them to probe and exploit vulnerabilities. By contrast, defenders in the open-source security community lack comparable, timely access to evaluate risks and develop countermeasures. The signatories contend that providing controlled, pre-release access to trusted researchers would strengthen defensive tooling and close a widening security gap.
Why It Matters for Crypto
Digital-asset platforms, wallets, and decentralized applications are frequent targets for phishing, malware, social engineering, and smart contract exploits. The organizations backing the letter say that improved collaboration between AI labs and independent security researchers would help harden critical crypto infrastructure and better protect users against emerging AI-enabled threats.
Signatories and Timing
The Bitcoin Policy Institute, a nonprofit research organization, published the letter on Aug. 10. More than 40 companies and groups across the Bitcoin and broader crypto ecosystem signed on, with Coinbase among the backers. The request centers on enabling vetted researchers to perform early-stage security testing of powerful AI systems under controlled conditions.
The signatories frame the proposal as a pragmatic extension of established security practices—such as coordinated vulnerability disclosure and structured red-teaming—adapted for frontier AI models. They argue this approach would allow defenders to anticipate and mitigate risks before new AI capabilities are widely deployed.