Yale Researchers Propose a Way to Keep AI Prompts From the Servers That Answer Them

A September preprint from six Yale computer scientists, two of them also with crypto research group IC3, describes Maverick, which hides a user's input from an outside AI server and checks its math, so far tested on one open-source model, Qwen3-4B.
The post Yale Researchers Propose a Way to Keep AI Prompts From the Servers That Answer Them appeared first on Unchained.
This is the summary Unchained publishes in its feed. The full article lives on their site.