Vitalik Buterin’s anonymous writing was successfully deanonymized by AI, which detected deeper reasoning patterns that persisted despite deliberate changes to his writing style.

Co-Investor’s analysis placed about a 20% confidence that Buterin authored the December 2024 EIP-7503 rewrite, roughly ten times higher than the next candidate among 27 documents.

Wang’s winning submission showed the author’s reasoning and explanations, not just prose, revealing Buterin as the author in Chinese with an English translation.

The development raises implications for crypto builders and contributors, highlighting that deanonymization can pose privacy and attribution challenges and is not universally reliable.

Privacy tech may need to address cognitive fingerprinting in addition to encryption and network anonymity to protect users in the AI era.

Consequently, privacy technologies should consider cognitive fingerprints, not solely traditional encryption or anonymity in the AI age.

AI is evolving from generating language to analyzing relationships between ideas, context, and logic, with wide-ranging impacts on academia, cybersecurity, misinformation analysis, and digital authentication.

This evolution expands AI’s role in understanding reasoning and context, affecting multiple sectors including open research, security, and digital identity.

The experiment marks a significant advance in AI’s ability to recognize deeper intellectual traits, influencing debates on privacy and identity in tech and crypto ecosystems.

This work aligns with broader research showing large language models can deanonymize at scale from unstructured text and echoes earlier attempts to attribute crypto authorship.

Online privacy could be increasingly at risk as AI tools may uncover authors across blogs, papers, social media, and forums, challenging anonymity for journalists, researchers, and whistleblowers.

The implications include heightened privacy risks since AI might reveal authorship regardless of wording changes across platforms.