Back to top
  • 공유 Share
  • 인쇄 Print
  • 글자크기 Font size
URL copied.

Coinbase Eyes Access to Anthropic's Restricted Claude Mythos AI for Cybersecurity Defense

Coinbase Eyes Access to Anthropic's Restricted Claude Mythos AI for Cybersecurity Defense. Source: Ivan Radic/Flickr(CC BY 4.0 Deed)

Coinbase is reportedly in talks with Anthropic to secure access to Claude Mythos Preview, a highly restricted frontier AI model built with advanced cybersecurity capabilities. The outreach, first reported by The Information, signals growing urgency among major crypto platforms to counter increasingly sophisticated AI-powered threats.

Anthropic launched Project Glasswing in early April 2026, a defensive cybersecurity initiative offering select partners limited access to the Mythos model. During testing, the system uncovered thousands of previously unknown zero-day vulnerabilities, including a 27-year-old flaw in OpenBSD and a 16-year-old bug in FFmpeg. Founding partners span some of the world's most influential technology and financial institutions, including Amazon Web Services, Apple, Google, JPMorgan Chase, Microsoft, and Palo Alto Networks. More than 40 additional organizations managing critical software infrastructure also gained access. Anthropic has further committed $100 million in compute credits and $4 million in funding directed toward open-source security groups participating in the program.

For Coinbase, the largest cryptocurrency exchange in the United States, the stakes are especially high. In 2025, the platform suffered a significant insider data breach that exposed personal information belonging to approximately 70,000 users after overseas support contractors were bribed by criminal actors. Coinbase refused to pay a $20 million ransom demand and instead offered a matching bounty for information leading to arrests. Separately, Anthropic's own research has demonstrated that AI agents are capable of autonomously exploiting smart contract vulnerabilities, simulating the theft of millions of dollars, a finding that reinforces why exchanges now view cutting-edge AI security tools as essential.

Claude Mythos will not be made publicly available. Anthropic plans to fold its core capabilities into future Claude model releases with additional safeguards. Post-preview pricing is set at $25 per million input tokens and $125 per million output tokens. Coinbase already uses Claude across customer support operations in over 100 regions worldwide. Whether the exchange ultimately secures formal Glasswing partnership status remains to be confirmed.

<Copyright ⓒ TokenPost, unauthorized reproduction and redistribution prohibited>

Most Popular

Comment 0

Comment tips

Great article. Requesting a follow-up. Excellent analysis.

0/1000

Comment tips

Great article. Requesting a follow-up. Excellent analysis.
1