How is agentic AI different from generative AI from a governance perspective?

Answer

Generative AI produces content — text, images, code. The governance concern is output quality and safety. Agentic AI takes actions — sending emails, executing trades, modifying databases, calling APIs. The governance concern shifts from 'did it say something wrong?' to 'did it do something wrong?' This is a fundamentally harder problem because actions have real-world consequences that can't be undone by regenerating a response.