Svmuu News: Researchers from the Ethereum Foundation’s Protocol Security Team stated in a blog post on Thursday that they have deployed a series of AI agents to test software that relies on Ethereum, searching for vulnerabilities in cryptographic systems, protocol code, and smart contracts. Vulnerabilities discovered by the AI agents included a remotely triggerable panic issue in `libp2p.gossipsub`, the peer-to-peer layer used by Ethereum consensus clients. This issue has been fixed and disclosed on GitHub as CVE-2026-34219. The researchers explained that the AI agents were organized into specialized roles—such as reconnaissance, search, gap-filling, and verification—to identify potential attack vectors, reproduce failures, and verify whether they apply to production code. The Ethereum Foundation stated that AI does not replace security researchers but rather changes the way they work, enabling the team to cover a much broader scope than manual reviews alone. However, researchers must exercise greater caution when evaluating a large number of seemingly credible findings. (Decrypt).