ETHERWOOD FOUNDATION: AI DISCOVERED LOOPHOLES THAT COULD LEAD TO THE VALIDATION OF NODES OFFLINE, BUT STILL NEEDS TO BE MANUALLY AUTHENTICATED

The ETA Foundation recently disclosed that its security team used AI agents to test software that operated at the ETA validation node and successfully discovered a loophole that could be triggered remotely and lead to a node collapse. However, the researchers stressed that manual clearance remains a key link in distinguishing between real loopholes and misstatement in the large number of security reports generated by AI. This discovered loophole was found in the Gossipsub protocol for the dissemination of information via the Taifung network, where the assailants could remotely trigger the node software into an abnormal computational state, causing the program to collapse and shut down, leaving the validation node offline until it was manually restarted by the operator, and the loophole was repaired and registered under the number “CVE-2026-34219). According to Nikos Bacevanis, a member of the security team of the Taifung Foundation Agreement, it was not AI's ability to detect loopholes that was truly surprising, but rather that the team spent a great deal of time identifying which loopholes were real and which were just seemingly credible “discoveries”。