Secure multi-agent system for privacy-preserving distributed computation

This invention describes a system that allows multiple artificial intelligence programs, called agents, to work together on tasks without revealing their sensitive data. It uses a special 'secure agent' within a protected computer environment to encrypt and decrypt the agents' internal information and manage how they securely pass data. Other agents perform specific parts of the task, and a 'coordinator agent' oversees their work, combines their encrypted results, and enforces security rules to keep everything private.