Anthropic has opened a new program that lets open-source projects receive security reports produced by its most capable AI models, Claude Mythos among them. Projects have to opt in by applying. Cointelegraph reported the launch on Oct 9, 2026.
Several crypto companies have already applied. Two of them are Nethermind, which develops an Ethereum client, and ZEUS, a Bitcoin and Lightning wallet. The report does not name the other applicants.
The report does not say which projects Anthropic will accept, when reports will start arriving, or what the reports will find. It also gives no account of any vulnerabilities found so far in Nethermind's or ZEUS's code.
For someone already holding a perpetual position on Bitget, Bybit, MEXC or OKX, this changes nothing directly. The program concerns code reviews for open-source software, and the story mentions no effect on funding rates, trading fees or order-book liquidity at any of these exchanges.
Source: cointelegraph — Crypto projects apply for Anthropic’s new frontier AI security scanner