Kraken's parent company Payward has joined Anthropic's Project Glasswing and gained access to Claude Mythos 5, the AI lab's frontier model purpose-built for finding and fixing software vulnerabilities. The Wyoming-based operator plans to deploy Mythos 5 across its environments in the coming weeks, with findings feeding into its existing security program. Payward also committed to disclosing vulnerabilities Mythos 5 surfaces in third-party open-source software back to project maintainers.
Why it matters
Project Glasswing launched in April 2026 with a tightly restricted cohort that already includes Amazon Web Services, Apple, Google, Microsoft, and JPMorganChase. Anthropic's stated rationale is dual: harden systems against AI-enabled attacks, and gather real-world data to build safeguards before a broader release. Payward argues crypto platforms face the same threat profile as other financial infrastructure, with around-the-clock operations, custody systems, and settlement rails that present lucrative targets.
The open-source disclosure angle is the part with ecosystem-wide reach. Vulnerabilities Payward surfaces in shared libraries will not only patch Kraken's stack; they will flow back to the maintainers who ship code the rest of crypto runs on.
Market impact
Payward co-CEO Arjun Sethi framed Mythos 5 as the first technology that meaningfully flips the defender-vs-attacker asymmetry crypto has lived with for years. "An attacker needs to find one flaw. A defender has to find all of them, first, every single day," Sethi said in the announcement. "Frontier AI is the first thing that flips that asymmetry."
For Kraken, the read is operational: faster vulnerability triage, lower cost per fix, and a stronger posture against AI-augmented phishing and exploit chains. For the broader exchange sector, Mythos 5 access through Glasswing is a signal that frontier defensive AI is starting to gatekeep who can credibly run a custody-grade venue.
Frequently asked questions
-
What is Anthropic's Project Glasswing?
Project Glasswing is an Anthropic initiative launched in April 2026 that gives selected firms access to restricted frontier AI models to help secure critical software and infrastructure. Initial cohort members include AWS, Apple, Google, Microsoft, and JPMorganChase.
-
What is Claude Mythos 5?
Claude Mythos 5 is Anthropic's most advanced defensive cybersecurity model, designed to analyze code at scale, find vulnerabilities, and help developers fix them. Access has initially been limited to organizations that operate or defend critical infrastructure.
-
Why is Kraken's parent Payward joining Glasswing?
Payward argues crypto platforms face the same threat profile as traditional financial infrastructure, with around-the-clock operations, custody systems and settlement rails that are lucrative targets for attackers. The company will deploy Mythos 5 across its environments and feed findings into its existing security…
-
Will Payward share findings with the broader crypto ecosystem?
Yes. Payward committed to disclosing vulnerabilities Mythos 5 surfaces in third-party open-source software back to the relevant project maintainers, potentially benefiting the wider crypto ecosystem that relies on the same shared libraries.
-
What did Payward's co-CEO say about the partnership?
Co-CEO Arjun Sethi framed Mythos 5 as the first technology that flips the long-standing defender-vs-attacker asymmetry in cybersecurity. "An attacker needs to find one flaw. A defender has to find all of them, first, every single day," Sethi said.
CoinDesk