The setup reflects his view that cloud-based AI tools pose a serious threat to personal privacy.
Crypto News
Ethereum co-founder Vitalik Buterin has shared the details of an AI system he runs entirely on his own hardware, describing it as both “private” and “secure” in a blog post published Wednesday. The setup reflects his view that cloud-based AI tools pose a serious threat to personal privacy.
Buterin runs an open-source model called Qwen3.5:35B using software called llama-server. He operates it on a laptop fitted with an Nvidia 5090 GPU. That configuration produces about 90 tokens per second, which he considers fast enough to use comfortably in daily tasks.
He keeps a full offline copy of Wikipedia and technical documentation stored locally on his machine. That reduces how often he needs to search the internet. He treats external search queries as a privacy risk because they reveal what users are looking for.
The AI system is also connected to his Ethereum wallet and messaging accounts. Buterin built and released a messaging tool that allows his AI to read Signal messages and emails without restriction. However, the tool blocks the AI from sending messages to anyone outside his own accounts unless a human manually approves each outbound action.
He said the principle applies equally to financial transactions. He recommended that any team building AI tools connected to Ethereum wallets cap autonomous spending at $100 per day. Any transaction above that threshold should require a human to confirm it before it goes through.
Buterin opened the post by citing a security finding that informed the design. Researchers found that roughly 15% of third-party tools built for OpenClaw, currently the fastest-growing repository on GitHub, contained malicious instructions. Some of those tools quietly sent user data to outside servers with no visible indication to the user.
He described his concern in direct terms, explaining that end-to-end encryption and locally run software had finally started to gain mainstream adoption, but he worries that cloud-based AI is now pushing that progress into reverse.
The AI guardrails follow the same logic as how he manages his crypto holdings. He stores 90% of his funds in a multisig Safe wallet. The keys are distributed among several trusted people, so no single individual can act alone. He said the AI setup is an extension of that same model applied to an automated context.
This article contains links to third-party websites or other content for information purposes only (“Third-Party Sites”). The Third-Party Sites are not under the control of CoinMarketCap, and CoinMarketCap is not responsible for the content of any Third-Party Site, including without limitation any link contained in a Third-Party Site, or any changes or updates to a Third-Party Site. CoinMarketCap is providing these links to you only as a convenience, and the inclusion of any link does not imply endorsement, approval or recommendation by CoinMarketCap of the site or any association with its operators. This article is intended to be used and must be used for informational purposes only. It is important to do your own research and analysis before making any material decisions related to any of the products or services described. This article is not intended as, and shall not be construed as, financial advice. The views and opinions expressed in this article are the author’s [company’s] own and do not necessarily reflect those of CoinMarketCap.
