The Double-Edged Sword of AI Peer Review: Why the First 'Double-Blind' Pilot May Decide the Soul of Academic Trust

In 2017, I spent four months auditing the smart contracts of EtherTrust, a popular but opaque ICO platform. I found a reentrancy vulnerability that could have drained $4.2 million in user funds. I published the exploit publicly, costing myself a lucrative consulting deal but earning something far more valuable: a reputation for putting conscience over consensus. Today, I read about the world's first massive-scale double-blind AI evaluation pilot for academic peer review, and I feel that same mix of hope and dread. The technology is dazzling. But I cannot shake the feeling that we are about to hand the keys to our intellectual integrity to a black box, without first asking who audits the auditor. This isn't about AI being good or bad. It's about whether we will let a machine define what knowledge is worth trusting, and whether the system that emerges will be more transparent than the one it replaces, or just a faster, more opaque version of the same human failings. Trust, after all, is earned, not mined.