Kraken parent Payward has joined Anthropic's Project Glasswing, according to reports from CoinDesk and Crypto Briefing, giving the crypto company access to Anthropic's defensive AI security program.

The move is notable because it brings a major digital-asset infrastructure operator into an initiative aimed at finding and fixing software vulnerabilities before attackers can exploit them. Anthropic's own Glasswing materials describe the program as a defensive deployment path for advanced cyber-capable models, with partners using the technology on critical codebases and security operations.

For Payward, the practical read is narrower than a product launch: this appears to be internal security infrastructure, not a new Kraken feature for customers. CoinDesk reported that Payward will use Claude Mythos 5 to help identify and repair vulnerabilities across software systems. Crypto Briefing described the same relationship as part of a push to scan crypto infrastructure for weaknesses.

That context matters for crypto because exchanges, custodians, payment platforms, and tokenization services remain high-value targets. AI can lower the cost of vulnerability discovery for defenders and attackers alike, making defensive review cycles more important for companies that handle private keys, trading systems, and settlement flows.

The verified takeaway is that Payward is adding Anthropic's Glasswing tooling to its security stack. The harder question is whether AI-assisted review can measurably reduce incident risk in production crypto systems, where integration quality, access controls, and operational discipline still matter as much as model capability.