Kraken Owner Payward Taps Anthropic’s AI for Enhanced Crypto Security Scanning

Key Highlights

• Kraken’s parent firm Payward has become a member of Anthropic’s Project Glasswing initiative, securing access to the Claude Mythos 5 AI cybersecurity tool.
• The crypto company intends to conduct comprehensive vulnerability scans across its entire software infrastructure in the upcoming weeks.
• Any confirmed security flaws discovered in open-source third-party software will be reported to the appropriate project maintainers.
• Since its April 2026 debut, Project Glasswing has grown to encompass approximately 150 organizations spanning over 15 nations.
• Access to Mythos 5 remains limited to approved organizations due to the technology’s potential dual-use nature in creating security exploits.

The parent entity behind cryptocurrency exchange Kraken, Payward, has officially joined Anthropic’s Project Glasswing security initiative. This partnership grants Payward access to Claude Mythos 5, representing Anthropic’s cutting-edge AI technology designed specifically for defensive cybersecurity operations.

The announcement came from Payward on August 17. According to the company’s plans, Mythos 5 will be implemented throughout its operational environments in the near future to perform comprehensive vulnerability assessments.

The Claude Mythos 5 system possesses the ability to analyze code at massive scale, detect security weaknesses, and recommend remediation solutions. Anthropic maintains strict controls over who can access this model since the same technology could potentially assist in developing functional exploits.

The infrastructure operated by Payward powers continuous digital asset trading, custody operations, and settlement functions that maintain 24/7 availability. According to the company, cryptocurrency platforms encounter security threats comparable to those facing other mission-critical financial systems.

In a statement, Payward co-CEO Arjun Sethi explained how the AI model shifts the advantage toward defenders. “A model can read every line of code the way an attacker would, at machine scale, so we find the flaw before anyone can build the exploit,” Sethi explained.

Results from vulnerability scans will be integrated into Payward’s established security assessment workflows. The company has not disclosed whether Mythos 5 will analyze live production environments or work with isolated code repositories.

Human Validation Remains Essential

According to Payward, all AI-identified security issues will undergo human verification before being confirmed as legitimate vulnerabilities. Complex software analysis by AI systems frequently generates false positive results.

The Ethereum Foundation arrived at comparable findings during its own AI-assisted security evaluations. Their research demonstrated that vulnerability reports produced by AI systems still require thorough human review, particularly when examining sophisticated protocol implementations.

Payward has also committed to sharing confirmed vulnerabilities found in open-source third-party software with the respective project maintainers. However, the company has not yet released details regarding its coordinated disclosure procedures or specific timelines for such reporting.

Understanding Project Glasswing

Project Glasswing was introduced by Anthropic in April 2026. Initial participants featured major technology and financial institutions including Amazon Web Services, Apple, Cisco, Google, JPMorganChase, and Microsoft.

The program has subsequently grown to include roughly 150 organizations distributed across more than 15 countries worldwide. Organizations seeking participation must satisfy stringent security criteria before obtaining access privileges.

According to reports, Anthropic’s earlier Mythos Preview version identified over 10,000 high-severity or critical vulnerabilities in commonly deployed software packages. The company’s public disclosure dashboard indicated a 90.8% accuracy rate for true positives among examined findings, although only 97 documented vulnerabilities had received upstream patches by May 2026.

Payward’s authorization followed a U.S. government determination permitting Mythos 5 distribution to entities that operate and protect critical infrastructure systems.

Organizations using Mythos 5 must agree to Anthropic’s 30-day data retention policy implemented for safety oversight purposes. Payward has not specified which code repositories or system data will be transmitted during vulnerability scanning operations.

The company has not announced specific performance metrics or objectives. Payward stated that scan results will supplement its current security framework rather than triggering automatic code modifications.

The post Kraken Owner Payward Taps Anthropic’s AI for Enhanced Crypto Security Scanning appeared first on Blockonomi.