Anthropic’s AI Can Automatically Detect Vulnerabilities—Coinbase Rushes to Buy It for Protection

Author: DLNews

Translation & Editing: TechFlow

TechFlow Intro: Anthropic’s new AI model, Claude Mythos, can autonomously discover and exploit zero-day vulnerabilities. The company considers it too dangerous for public release and has restricted access to only a handful of major technology partners. Coinbase and Binance are racing to acquire it to bolster their defenses—but this leaves the rest of the industry completely exposed to AI-powered attacks. In 2025, the crypto industry has already lost $3.4 billion to hackers, and AI can identify vulnerabilities that traditional audits entirely miss—in seconds.

Major cryptocurrency exchanges are competing for access to Anthropic’s Claude Mythos, a new AI model whose purported hacking capabilities have led the company to restrict its availability to only a select group of large technology partners.

According to a report by The Information on April 14, Coinbase and Binance are vying to purchase access to this new model to strengthen their defenses.

The urgency stems from Claude Mythos’s alleged ability to autonomously discover and exploit zero-day vulnerabilities—security flaws attackers can exploit before developers even know they exist—across major operating systems and web browsers. Worse still, Anthropic claims the model is so dangerous that the company has not yet set a public release date.

Yet while the largest exchanges rush to arm themselves with AI-driven defenses, the rest of the crypto ecosystem is locked out—and left vulnerable.

For Deddy Lavid, CEO of cybersecurity firm Cyvers Alert, this scenario is a nightmare.

“If AI can identify vulnerabilities in core internet infrastructure at scale, crypto will be among the first markets to feel the impact,” Lavid told DL News.

“Our ecosystem runs on browsers, wallets, APIs, open-source libraries, and development tools—many of which are directly connected to value transfer.” And the potential scale of damage is staggering.

“Hundreds of millions to billions of dollars,” he said.

According to blockchain analytics firm Chainalysis, the crypto industry has already suffered $3.4 billion in losses to hackers and malicious actors in 2025.

Indeed, for an industry built on irreversible transactions and lean security teams, the risk is existential. This is because AI does not merely read code faster than humans—it rapidly explores vast amounts of code to uncover emerging vulnerabilities that traditional audits completely overlook.

Combine this capability with crypto’s instant finality and fragmented infrastructure, and the gap between vulnerability discovery and catastrophic loss shrinks from weeks to seconds.

No Layer Is Immune

Data from Cyvers shows that large-scale historical losses have struck every layer of the crypto trading stack.

Centralized exchanges lost nearly $2 billion, decentralized networks were attacked for over $500 million, cross-chain bridges were breached for nearly $1 billion, and user wallets became targets, Levid said.

AI threats can severely impact crypto in three primary ways: small teams, long-unupdated codebases, or abandoned projects.

“We frequently see protocols safeguarding hundreds of millions of dollars maintained by lean teams, aging codebases, or partially abandoned projects,” Levid said. “In an AI-driven threat environment, this gap becomes a major risk amplifier.”

What about Bitcoin—the dominant crypto protocol?

“Bitcoin’s core protocol is relatively simple and battle-tested, far less exposed to complex logic errors than other ecosystems,” Levid told DL News. “AI is unlikely to suddenly ‘break’ Bitcoin itself.”

However, the infrastructure surrounding this top cryptocurrency—wallets, exchanges, L2s—remains vulnerable.

“AI can discover vulnerabilities at these layers, and since this is where most users interact with Bitcoin, actual losses will occur—not in the protocol itself, but at the edges,” Levid said.