DeFi users have been warned that approvals will be revoked ahead of the launch of Anthropic’s Mythos AI

Token approvals often accumulate unnoticed over time, creating potential exposure if previously trusted contracts are compromised.

Anthropic is reportedly set to launch a public version of its Mythos AI model, and cryptocurrency analyst The DeFi Investor is urging DeFi users to act before that happens.

The concern is based on how good Mythos is at finding vulnerabilities in software, and a more widely accessible version of it could speed up how quickly attackers discover and exploit vulnerabilities in DeFi protocols.

What the DeFi community needs to do

In a June 9 post on X, The DeFi Investor It is recommended Followers should revoke all token approvals, only use highly vetted decentralized applications, and distribute funds across multiple wallets to reduce single points of failure.

For those who are not familiar, token approvals are permissions that users grant to smart contracts, allowing the contracts to spend tokens on their behalf. They tend to accumulate silently over time, and represent a constant attack surface if any agreed contract later turns out to be compromised.

“The scary thing about Mythos is that it is insanely good at finding extreme vulnerabilities,” The DeFi Investor wrote. “Claude Opus 4.8 recently identified a serious bug in Zcash, and Mythos is supposed to be better than Opus 4.8.”

They added that DeFi will face a huge stress test in the next few months, and indeed, the Zcash vulnerability they mentioned gave a concrete example of this.

The privacy coin lost more than 35% of its value in a single day after a security researcher using artificial intelligence discovered a flaw in its protected Orchard pool that would have allowed bad actors to mint new ZEC tokens indefinitely. Major cryptocurrency investor Arthur Hayes has witnessed exit His entire stance on ZEC, with uncertainty growing as to whether anyone has actually exploited the flaw.

The Mythos have since April been restricted to about 50 organizations, including Amazon, Apple, Google and Microsoft, through a humanitarian initiative known as Project Glasswing, in an attempt to put the model’s working capabilities to use for defensive purposes. According to Bloomberg, Anthropic plans to Expand Which revolves around 150 other organizations in 15 countries.

You may also like:

However, this has been stated by multiple sources, including the Terrorist Financing Targeting Center and journalist Alex Heath He claimed That the public version of Mythos will have “significant guardrails” and will not be as lenient as what Project Glasswing partners have access to.

There has been discussion about DeFi already

The security advice from the DeFi Investor came at a time when the conversation was revolving around the viability of decentralized finance.

In late May, OpenZeppelin co-founder Manuel Arauz Announce ‘All DeFi is unsafe’ He said he advised people to exit positions in major protocols, including Aave, MakerDAO and Compound. The reason he did this was because AI had tilted the balance of security so far towards attackers that no protocol could currently be trusted to hold users’ funds securely.

In fact, several cryptocurrency projects have been hit in the past few months, including attacks on KelpDAO and Drift Protocol in April, resulting in a loss of more than $570 million combined. Recently, it was reported that hackers It has been withdrawn At least $30 million worth of Humanity Protocol’s H token from 17 wallets.

However, according to Aave Chan Initiative founder Mark Zeller, concerns about AI are overblown, as less than 10% of DeFi security failures last year were caused by code-level vulnerabilities.

Anthropic’s own position, according to Bloomberg, is that in the long term, AI will favor defenders, but “the transition period will be risky.”