[CLSTR]
< Back to all clusters
[TECHNOLOGY] · 4 sources

started · updated

Vitalik Buterin tests privacy-focused AI for personal health data

Vitalik Buterin privacy-preserving AI zkAPI local LLM data privacy

Ethereum co-founder Vitalik Buterin is conducting a self-experiment to test a privacy-focused artificial intelligence setup. The goal is to use personal health and travel data to generate customized diet and exercise recommendations while preventing sensitive information from leaking to remote, high-capacity AI models.

The architecture relies on a three-layer privacy strategy. A local model, specifically Alibaba’s Qwen3.8-Flash-Next, acts as the primary controller. This local model rewrites user prompts to strip away identifying details and personal context before sending necessary queries to more powerful remote models. To further secure the process, the setup utilizes zkAPI to decouple payment identities from model requests and Tor to mask internet traffic and IP information.

While the experiment has proven functional, Buterin noted a trade-off between privacy and utility: as the system becomes more cautious about the data it shares, the quality and depth of assistance from remote models can decrease. Additionally, current tests face technical hurdles, such as Tor latency being 10 to 100 times higher than standard speeds and local processing speeds currently running at 20–30 tokens per second, whereas Buterin aims for over 100 tokens per second for a smoother experience.

Entities

Alibaba · Ethereum · Qwen · Tor · Vitalik Buterin

Claims

What the coverage asserts, and how many sources carry each claim.

  • [○ 1 SOURCE] The local Qwen model runs at approximately 20 to 30 tokens per second. crypto.news
  • [○ 1 SOURCE] The system uses zkAPI to separate payment identity from model requests. crypto.news
  • [● 2 SOURCES] Vitalik Buterin is conducting a self-experiment using personal health and travel data to generate personalized diet and exercise recommendations. coin-turk.com · crypto.news
  • [○ 1 SOURCE] Tor latency in the current tests is 10 to 100 times higher than standard speeds. crypto.news
  • [○ 1 SOURCE] Tor is used within the setup to mask IP information. crypto.news
  • [○ 1 SOURCE] The experiment employs Alibaba’s Qwen3.8-Flash-Next as the local model. crypto.news
  • [○ 1 SOURCE] Increased privacy in the setup can limit the helpfulness of the remote models. coin-turk.com
  • [● 2 SOURCES] The privacy-focused AI setup utilizes a local model to manage queries and limit data exposure to remote models. coin-turk.com · crypto.news

Sources

Vitalik Buterin, yapay zeka deneyinde gizlilik odaklı yöntemi açıkladı [coin-turk.com]
Vitalik Buterin tries AI that keeps personal data private [crypto.news]
Vitalik Buterin Tests AI Privacy Through zkAPI and Tor Routing [blockonomi.com]
Ethereum's zkAPI Gives AI Payments a Privacy Cloak With a Catch [news.bitcoin.com]