Vitalik Buterin has confirmed a winner in his AI challenge to unmask his anonymous writing, after researcher Franklyn Wang traced a hidden EIP-7503 revision back to the Ethereum co-founder.

Buterin launched the experiment on June 22, offering up part of his own anonymity to test whether AI stylometry can identify hidden authors. For 13 days, no one succeeded.

AI Challenge Winner Matched Reasoning, Not Prose

Wang ran the search through Co-Invest, an AI research engine, and flagged a December 2024 revision of EIP-7503, the Zero-Knowledge Wormholes privacy proposal. A throwaway account submitted the rewrite, which now forms roughly 75% of the proposal’s text.

The disguise nearly held. Keyvan Kambakhsh, an original EIP-7503 author, reviewed and approved the anonymous edit at the time. According to Wang’s analysis, his model gave the pick just 20% confidence, yet 10 times its next candidate. The search reportedly took about two hours.

Follow us on X to get the latest news as it happens 

Buterin drafted the revision in Chinese, translated it locally with Alibaba’s Qwen2.5 model, and manually fixed the output. However, that camouflage covered only his prose.

“Notice that the stylistic hints that his AI picked up on were intellectual habits and style of math and algorithm explanation, which bypassed my obfuscation strategy (which only covered prose) completely.”

Buterin confirmed the result on Monday. Wang, for his part, argued the same engine could hunt trading signals across news and on-chain data.

What the Result Means for Pseudonymous Developers

Stylometry has unmasked authors before. In 2013, forensic linguists identified J.K. Rowling as crime novelist Robert Galbraith through vocabulary and phrasing. Buterin’s test suggests detection now reaches deeper, into how an author reasons rather than how they write.

That shift matters for an industry built on pseudonyms, from Satoshi Nakamoto down. Ethereum alone recently passed 1 million developers, while European regulators already fuel crypto privacy fears.

Buterin has championed privacy for years, from co-authoring the Privacy Pools paper to his Lean Ethereum roadmap. His self-experiment also sharpens the debate over AI safety rules as models grow more capable.

Whether obfuscation can catch up, or whether this “thoughtprint” detection keeps improving, may become clearer as researchers rerun Wang’s method on other anonymous work.

Источник: BeInCrypto