Anthropic has opened enrollment for its OSS Scanner, an artificial-intelligence-based service that automatically analyses open-source repositories and returns vulnerability reports. The move arrives as crypto developers seek stronger defensive tools against increasingly sophisticated attackers who are beginning to exploit frontier AI models.
AI-driven scanner for open-source code
Anthropic announced the OSS Scanner on Thursday, positioning it as an upgrade to its earlier Project Glasswing effort. While the company already performs manual scans and issues reports after human review, the process can be slow. By leveraging its most powerful models, including Claude Mythos, the new scanner promises near-real-time feedback once code is examined. Participation is voluntary, and projects are evaluated individually based on factors such as their role in critical infrastructure, exposure to remote threats, and the size of their user base.
Crypto projects line up for early access
Within hours of the program’s launch, a handful of blockchain-related teams opened pull requests on the scanner’s GitHub repository to request audits. Nethermind, a developer of an Ethereum client, asked for a comprehensive review of its entire code repository. ZEUS, a self-custodial wallet supporting Bitcoin and Lightning Network transactions, requested an assessment focused on payment handling, private-key protection, and Lightning connectivity. Other applicants include VirtEngine, a decentralized cloud-computing marketplace built on the Cosmos SDK, as well as teams working on AI assistants, agent-security tools, machine-learning infrastructure, development utilities, cloud storage, and energy-system controls.
Context of rising AI-assisted threats
The crypto sector has already reported incidents where attackers appear to be using AI to accelerate exploit development. Earlier this year, Bitcoin swap service Boltz halted operations, citing a pace of attacks that outstripped its ability to patch. Similarly, the crypto-payment platform PayPerQ has experienced repeated breaches it suspects are AI-powered. Anthropic warned that, in the short term, AI may tip the balance toward attackers because exploiting vulnerabilities becomes cheaper, while verification and remediation remain labor-intensive.
Potential impact on ecosystem security
If the OSS Scanner delivers timely, high-quality vulnerability reports, participating projects could close gaps before they are weaponised. Faster disclosure may also reduce the window of opportunity for malicious actors who rely on delayed human review cycles. By prioritising projects with significant user impact, Anthropic hopes to amplify defensive benefits across the broader blockchain infrastructure.
Why it matters
The integration of cutting-edge AI into open-source security tooling marks a shift in how blockchain projects protect their code. As attackers adopt similar technologies, defenders must match that pace to preserve trust and operational stability. Anthropic’s OSS Scanner offers a scalable, model-driven approach that could become a benchmark for open-source security in the crypto space, especially for critical components like Ethereum clients and Bitcoin wallets.
