What I Found Trying to Build FHE from One-Time Pads

236 · · March 27, 2026, 9:01 p.m.
Summary
The blog post discusses the risks of using LLM providers for data processing, emphasizing potential misappropriation of data and highlighting options for running local models, albeit at high hardware costs. It raises concerns about trust in providers like OpenAI and Anthropic, linking them to issues of privacy and government surveillance.