Ethereum client developer Nethermind and self-custodial Bitcoin and Lightning wallet ZEUS are among several crypto companies that have applied to a newly launched Anthropic program giving open-source projects access to security reports from its most powerful AI models.

The projects submitted enrollment requests on Friday for OSS Scanner, a service Anthropic launched on Thursday to help open-source developers find and fix vulnerabilities before attackers exploit them.

“These reports will be generated by our strongest models (including Claude Mythos) to give open-source projects the largest defensive advantage,” Anthropic said in its announcement.

The applications come as crypto firms push for access to frontier AI models to defend against increasingly capable attackers. Cybersecurity experts have warned that uneven access could leave defenders at a disadvantage as powerful alternatives become more widely available.

Opt-in service builds on Project Glasswing

Anthropic said OSS Scanner is an opt-in service that follows its work on Project Glasswing. The company already regularly scans open-source software for vulnerabilities and sends reports after human review, but said the manual review process is slow, “so we're not always able to share vulnerabilities as quickly as we would like.”

OSS Scanner will instead deliver vulnerability reports to participating projects as soon as their code has been scanned. For accepted projects, that shortens the interval between a scan flagging a weakness and the maintainers hearing about it.

Crypto applicants

Pull requests to the OSS Scanner GitHub repository show Nethermind has applied for audits of its entire repository, while ZEUS has asked for its app to be examined for weaknesses affecting payments, private keys and connections to Lightning services in a separate request. Another applicant is VirtEngine, a decentralized cloud computing marketplace structured as a Cosmos SDK chain, which has also submitted a request.

Other applicants include developers of AI assistants, agent-security tools and machine-learning infrastructure, alongside projects providing software development tools, cloud storage and energy-system controls. None of the pull requests had been merged at the time of publication.

Anthropic said projects will be assessed on a case-by-case basis, taking into account their importance to infrastructure and user security, their exposure to remote attacks and how many users or other projects depend on them. By those measures, client software like Nethermind's sits close to the core of Ethereum transaction processing, while a self-custodial wallet like ZEUS leaves users with no intermediary to absorb the impact of a key or payment flaw. Whether the outstanding pull requests are merged will show how readily crypto projects gain that access in practice.

AI-assisted attacks on the rise

Crypto companies have reported AI-assisted attacks this year. Bitcoin swap provider Boltz suspended operations in August, saying attackers were developing exploits faster than its team could patch them, while crypto-payment service PayPerQ reported repeated attacks it suspected were AI-powered.

Anthropic warned on Thursday that AI may favor attackers in the near term, as exploitation becomes cheaper while verifying and fixing vulnerabilities remains slow and dependent on people.
Related: Vitalik Buterin backs crypto 'bunker mode' amid rapid AI math advances