Kraken parent Payward joins Anthropic’s Project Glasswing for AI security push

Payward, the parent company of Kraken, has joined Anthropic’s Project Glasswing to leverage Claude Mythos 5, an AI model designed for identifying and fixing software vulnerabilities. Project Glasswing is Anthropic’s program that grants selected companies access to restricted AI models for enhancing software and infrastructure security against cyber threats.

Other participants in the project include Amazon Web Services, Apple, Google, Microsoft, and JPMorgan Chase. Payward, based in Wyoming, intends to deploy Mythos 5 across its systems within the next few weeks, with the results contributing to its existing security measures. Moreover, the company intends to share discovered vulnerabilities in third-party open-source software with the respective project maintainers.

This move follows the U.S. government's decision to make Mythos 5 available to U.S. organizations responsible for critical infrastructure. Anthropic introduced Project Glasswing in April 2026 and has since expanded it to include firms in the tech and finance sectors. As AI poses an increasing risk to crypto companies, attackers gain the ability to locate vulnerabilities more rapidly, automate attacks, and craft more convincing phishing and social-engineering campaigns, intensifying the urgency for the sector to deploy defensive technologies.

Claude Mythos 5, Anthropic’s most advanced model for defensive cybersecurity, is specifically crafted to analyze code on a large scale, pinpoint vulnerabilities, and assist developers in rectifying them. Initially, access to this model has been restricted to organizations operating or defending critical infrastructure, as Anthropic works on measures to broaden its deployment.

Payward contends that cryptocurrency platforms encounter security challenges comparable to other essential financial infrastructure. These platforms operate continuously and can serve as attractive targets for cybercriminals. "Security has always been an uneven battle. An attacker must discover a single flaw; a defender must uncover all of them, every single day," stated Payward co-CEO Arjun Sethi in a company release.

"Frontier AI flips this imbalance." Anthropic has announced plans to widen access to Mythos-class cybersecurity capabilities as it develops safeguards for broader utilization.