Payward Inc., the parent company of cryptocurrency exchange Kraken, has joined Anthropic’s Project Glasswing, an initiative aimed at using artificial intelligence to identify and address cybersecurity vulnerabilities.
The partnership will see Payward deploy Anthropic’s specialised cybersecurity AI model to scan for weaknesses across its infrastructure. Findings from the project will be shared as open source contributions, according to a statement released Sunday.
Project Glasswing was launched by Anthropic, the AI research company behind the Claude large language model, to bring together technology firms, security researchers, and AI developers in a coordinated effort to improve digital security. The initiative focuses on using advanced AI systems to detect vulnerabilities before they can be exploited by malicious actors.
Payward’s participation marks the first major involvement by a cryptocurrency exchange operator in the project. The company operates Kraken, one of the world’s largest digital asset trading platforms by volume.
Open source vulnerability data
Under the terms of the partnership, Payward will use Anthropic’s cybersecurity focused AI model to conduct internal security assessments. Any vulnerabilities discovered through this process that have broader industry relevance will be shared publicly as open source intelligence.
The move reflects growing interest across the cryptocurrency industry in using AI powered tools to address security challenges. Digital asset platforms have long been prime targets for hackers, with billions of dollars lost to breaches and exploits over the past decade.
Anthropic has positioned Project Glasswing as a collaborative defence mechanism, arguing that widespread adoption of AI security tools can raise baseline protections across the technology sector. The company has not disclosed how many organisations are currently participating in the initiative.
The cryptocurrency sector has faced persistent security threats, ranging from smart contract exploits to phishing attacks and insider breaches. Exchanges in particular face scrutiny from regulators and customers over their ability to safeguard digital assets.
Payward’s decision to join the AI security project comes as regulatory pressure mounts globally on crypto firms to demonstrate robust operational controls. Several jurisdictions, including the European Union and Singapore, have introduced licensing frameworks that include stringent cybersecurity requirements.
Kraken has not experienced a major security breach in recent years, but the broader industry continues to grapple with high profile incidents. In 2024 alone, more than $2 billion was stolen from decentralised finance protocols and centralised exchanges, according to blockchain security firm CertiK.
Anthropic has not disclosed technical details about how its cybersecurity AI model operates, citing security concerns. The company said the model has been trained specifically to identify common vulnerability patterns and assist human security teams in threat assessment.
Payward did not specify a timeline for how long it would participate in Project Glasswing or what categories of vulnerabilities it expects to prioritise in its security sweeps.