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

Kraken parent Payward has joined Anthropic’s Project Glasswing, giving the crypto company access to Claude Mythos 5, an artificial intelligence model designed to find and fix software vulnerabilities.

Project Glasswing is Anthropic’s initiative that allows selected firms to use restricted frontier AI models to help secure critical software and infrastructure against cyber threats. The project already includes tech giants such as Amazon Web Services, Apple, Google, and Microsoft, as well as financial institutions such as JPMorganChase.

The Wyoming-based company said Monday it plans to deploy Mythos 5 across its environments in the coming weeks, with findings feeding into its existing security program. Payward also plans to share vulnerabilities found in third-party open-source software with the relevant project maintainers.

The move comes after the U.S. government made Mythos 5 available to U.S. organizations that operate and defend critical infrastructure, according to Payward. Anthropic launched Project Glasswing in April 2026 and has since expanded the initiative to organizations in technology and finance.

AI is also emerging as a growing threat to crypto companies, giving attackers tools to find vulnerabilities faster, automate attacks and make phishing and social-engineering campaigns more convincing. That raises the stakes for an industry already a frequent target for hackers, while fueling a race to deploy the same technology on the defensive side.

Claude Mythos 5 is Anthropic’s most advanced model for defensive cybersecurity, designed to analyze code at scale, identify vulnerabilities and help developers fix them. Access has initially been limited to organizations that operate or defend critical infrastructure, as Anthropic works on safeguards for a broader rollout.

Payward argued that crypto platforms face security challenges similar to other critical financial infrastructure. Exchanges, custody systems and settlement rails operate around the clock and can present lucrative targets for attackers.

“Security has always been an unfair game. An attacker needs to find one flaw. A defender has to find all of them, first, every single day,” Payward co-CEO Arjun Sethi said in the release. “Frontier AI is the first thing that flips that asymmetry.”

Anthropic has said it plans to expand access to Mythos-class cybersecurity capabilities as it develops safeguards for wider use.

Source