Kraken parent company Payward has announced that it is joining Anthropic’s Project Glasswing initiative, a program designed to use cutting-edge artificial intelligence models to strengthen the security of critical software infrastructure. Payward’s security team is integrating the latest model, Claude Mythos 5, into its defensive cybersecurity operations to automate scanning of enterprise environments and identify potential vulnerabilities.
Project Glasswing was launched in April 2026, when Anthropic discovered that high-end AI models can outperform almost all humans at identifying and exploiting software vulnerabilities. Since then, partners, including Payward, have discovered thousands of high or critical security vulnerabilities. This initiative not only helps companies protect their infrastructure, but also contributes to the security of the broader crypto ecosystem by reporting the vulnerabilities found to the appropriate open source project maintainers.
Claude Mythos 5 as a new line of defense
Claude Mythos 5 is a cybersecurity model developed by Anthropic that is designed to read every line of code as a potential attacker would, but at machine scale. Payward Co-CEO Arjun Sethi emphasized that „frontier AI is the first factor that reverses the asymmetry“ – traditionally, an attacker had the advantage because they only needed to find one vulnerability, while a defender had to find all of them. The model allows the company to find flaws before others can create exploit code.
Payward plans to integrate Claude Mythos 5’s vulnerability detection into its existing three-tier triage and remediation process. This means that the model’s results will be automatically fed to security teams – both red (penetrating) and blue (defensive) – for review and rapid resolution. The company also maintains ISO 27001 and SOC 2 certifications, and this AI tool will complement existing security measures.
Open source component security
An important aspect of Project Glasswing is responsible disclosure of open source vulnerabilities. Payward has committed to reporting vulnerabilities found in open source components to the project maintainers so that they can quickly fix the code. This not only protects Payward’s infrastructure, but also contributes to the security of the broader crypto community, as many platforms use the same open source foundation.
Other partners and sector context
Decrypt reports that Payward is the first crypto company to join Project Glasswing and gain access to Claude Mythos 5. This follows an open letter sent by more than 40 Bitcoin and crypto companies, including Kraken, demanding that AI labs give trusted defenders access to their most powerful models. The letter, organized by the Bitcoin Policy Institute, emphasized that financial infrastructure developers must have tools to find and fix vulnerabilities before attackers themselves.
Payward's approach to future security
Payward says the project will help strengthen security in the infrastructure that serves millions of customers around the world, while also contributing to security improvements for the broader crypto ecosystem. The company emphasizes that AI models like Claude Mythos 5 provide the ability to read code at „machine scale,“ allowing them to find vulnerabilities faster than traditional methods.
Conclusions
Payward’s joining of Anthropic Project Glasswing and the use of Claude Mythos 5 reflects the growing role of AI in cybersecurity. This initiative not only gives the company a powerful tool for vulnerability detection, but also promotes a shared security culture by sharing the flaws found with the open source community. Further development of AI models and their integration into real-world security processes can become standard practice as companies strive to maintain an edge against increasingly resourceful cyber threat sources.
—
We invite you to contact Krikis IT if you would like to learn how artificial intelligence can help automate your company's cybersecurity processes.
We invite you to contact Krikis IT if you would like to learn how artificial intelligence can help automate your company's cybersecurity processes.
Sources
- Crypto Briefing – Payward joins Anthropic's Project Glasswing to hunt software vulnerabilities with Claude Mythos 5
- Decrypt - Kraken Parent Payward Joins Glasswing, Gets Access to Claude Mythos to Hunt Security Flaws
- Biztoc.com - Kraken parent Payward joins Anthropic's Project Glasswing for AI security push
