The Ethereum Foundation has turned AI agents loose on the Ethereum network to find security flaws before malicious actors can exploit them. The Protocol Security team is running coordinated AI agents that probe critical infrastructure for vulnerabilities, shifting the core security workload from discovery to validation.
According to the Foundation, the AI agents do find real bugs, but the vast majority of alerts are false positives. This means researchers must now spend more time proving which vulnerabilities are genuine rather than relying on the AI to deliver clean results. The approach represents a significant change in how Ethereum approaches security testing.
Notably, the deployment comes as blockchain security remains a top concern across the crypto sector. While automated vulnerability scanning is common, using AI agents at the protocol level is a relatively new strategy that could set a precedent for how Layer-1 networks handle security auditing going forward.
For Ethereum's market position, the move signals to investors and developers that the Foundation is proactively addressing security risks. This could help maintain confidence in the network's long-term reliability, though the high false positive rate suggests the technology is still maturing.
Critics argue that relying on AI for security testing may create a false sense of safety, especially given the high rate of false positives. Some security researchers warn that AI agents may miss novel attack vectors that human auditors would catch, potentially leaving the network exposed to sophisticated exploits.