Ethereum co-founder Vitalek Boutherin confirmed that AI analysis of Co-Invest head Franklin Van correctly identified the authorship of his anonymous redraft EIP-7503 dated December 2024. Boutherin launched a public experiment in June 2026 to test the threat of AI for online anonymity by suggesting that participants find a document he published anonymously.

The winner was Wang, whose AI team analyzed 27 texts and estimated that Buterin’s authorship was ~20% - about 10 times higher than that of the next candidate. Boutherin purposefully tried to hide the authorship: he wrote the text in Chinese, translated it through Qwen 2.5 and manually fixed the errors. However, AI recognized him by the peculiarities of explaining mathematical and technical ideas rather than linguistic markers.

The experiment was inspired by a February report from researchers at ETH Zurich and Anthropic about the ability of LLM to help de-anonymize users. Van previously used a similar approach to try to identify Satoshi Nakamoto, but without much confidence. Boutherin noted: the style traits that AI discovered were intellectual habits and reasoning - not vocabulary or grammar.