Crypto Projects Tap Anthropic's AI Scanner for Security Audits

Several crypto companies, including Ethereum client developer Nethermind and Bitcoin Lightning wallet ZEUS, have applied for Anthropic's newly launched OSS Scanner, an opt-in service that delivers AI-generated vulnerability reports to open-source projects. The program, announced Thursday, leverages Anthropic's most powerful models—including Claude Mythos—to identify security weaknesses in code before attackers can exploit them. Projects submitted enrollment requests via pull requests to the OSS Scanner GitHub repository on Friday, none of which had been merged at the time of publication.

Nethermind applied for comprehensive audits of its entire repository, while ZEUS requested scrutiny of its self-custodial Bitcoin and Lightning wallet, specifically targeting vulnerabilities in payments, private key management, and Lightning Service connections. Decentralized cloud computing marketplace VirtEngine, built as a Cosmos SDK chain, also submitted an application. Other applicants span AI assistants, agent-security tools, machine-learning infrastructure, software development platforms, cloud storage, and energy-system controls—reflecting broad interest across the open-source ecosystem.

Anthropic said the scanner builds on its Project Glasswing initiative and addresses a key bottleneck in its existing vulnerability disclosure workflow: the slow pace of human review. With OSS Scanner, participating projects will receive reports as soon as their code has been scanned, rather than waiting for manual triage. Acceptance will be assessed case by case, with Anthropic weighing factors such as infrastructure importance, exposure to remote attacks, and downstream dependency risk. The move comes as crypto firms increasingly seek access to frontier AI models to counter increasingly capable attackers. Bitcoin swap provider Boltz suspended operations in August, citing that exploit developers were outpacing its team's ability to deploy patches.