Recently, some friends have been asking whether Mythos will have a significant impact on the encryption field and Ethereum? Setting aside clickbait headlines or traffic articles, Mythos has a tangible impact on the field of cryptography and Ethereum. We need to develop new defense methods. Simply put, Anthropic's newly released Claude Mythos Preview is a super powerful model, so powerful that it is not yet fully open for release. It has a characteristic of being good at finding software bugs. In just a few weeks, it has identified thousands of super serious zero day vulnerabilities (bugs that no one had previously discovered) on its own. These bugs are hidden in: • Mainstream operating systems (Windows, Linux, macOS, etc.) • Mainstream browsers Even some old code that has been used for ten or twenty years (a 27 year old bug was also dug out by it) What's even more headache inducing is that, It can also write code on how to attack. What does this mean for the crypto ecosystem and Ethereum? The most concerning issue for everyone is safety. Since it can find vulnerabilities so well, is it worth paying attention to in the field of encryption? It is indeed worthy of attention. The field of encryption relies on two things to protect user wallets: One is to rely on cryptographic algorithms (mathematically secure, such as signatures and encryption), The second is the running software and code, such as nodes/wallets/smart contracts/browsers/operating systems, etc. Mythos has little impact on cryptographic algorithms, but what has a significant impact on cryptographic algorithms is that quantum computers, which have been widely discussed in the community a few days ago, are a type of threat. However, in the short term, it is still okay. The main impact this time is the second point: The nodes, DeFi protocol, wallets, and bridges (cross chain things) of Ethereum run on ordinary computers Linux、 On the browser. Once the smart contract code is deployed, it cannot be modified, and if there are hidden bugs inside, the money inside may be stolen. Previously, finding bugs was quite troublesome, requiring manual auditing and various tools. Relatively speaking, it was expensive, slow, and easy to miss bugs. Mythos has the ability to automatically identify bugs and write attack code at a low cost. If this level of AI capability spreads in the future and is acquired by attackers, the attack capability will increase and the attack cost will decrease. This poses significant risks for DeFi, especially for potential vulnerabilities in old contracts and bridges. The probability of nodes being remotely controlled and RPC being attacked has also increased. Of course, there is also a positive side. Anthropic has not yet made Mythos public. Instead, they have launched a "Project Glasswing" program, which is currently only available to some large companies and critical infrastructure teams (including Linux, Google, Microsoft, Apple, etc.). Use it first Help everyone fix bugs and reinforce open source code and critical software first. In the short term, there is still a certain window of opportunity within approximately one year. For now, Mythos is only used by a few large companies. If the Ethereum ecosystem can leverage its partners to scan code with Mythos, there is a chance to do defense in advance. But in the long run, developers need to adopt "AI formal verification+autonomous red team" as a necessary pre deployment step to reduce the risk of launching new contracts. Ultimately, Ethereum and the crypto ecosystem need to promote "formal verification+multi client+top-level AI defense monitoring", which seems far more urgent than "quantum computing".