Building policy enforcement into the AI stack with Praxis
IBM Research
IBM Research introduced Praxis, a programmable policy engine designed as a security guard for AI agents.
IBM Research has introduced the Praxis Policy Engine, designed to build policy enforcement directly into the artificial intelligence stack. Operating from inference through gateways, agents, and policy controls, Praxis functions as a programmable security guard for AI agents. The tool aims to help manage security and governance within AI architectures.
Why it matters: It provides a mechanism for embedding security and policy enforcement into the AI stack to secure AI agents.

