Ethereum finds AI bug but warns limits
Share
- The Ethereum Foundation said AI agents found a real software vulnerability but still require human verification.
- The confirmed flaw affected libp2p and was later disclosed as CVE-2026-34219.
- The Foundation said AI works best as a security assistant rather than a replacement for researchers.
The Ethereum Foundation said AI agents found a real vulnerability in software used by Ethereum (CRYPTO:ETH), but human experts were still needed to confirm the results.
The confirmed bug affected the gossipsub component of libp2p and was fixed before being disclosed as CVE-2026-34219.
The Foundation said most AI reports were false positives, duplicates or issues that could not be reproduced on production code.
The Foundation said every accepted report must include a reproducible test case, clear evidence and a defined security failure before it is treated as a real vulnerability.
The organisation uses several AI agents to inspect code, test security ideas, remove duplicate findings and independently validate potential bugs.
The update follows the Ethereum Foundation's June restructuring, which reduced its workforce by about 20% while focusing resources on long-term Ethereum development.
At the time of reporting, Ethereum price was $1,744.20.
