AI agents face a harder question: who is responsible when they act?

As AI systems gain the ability to take actions instead of simply answering questions, responsibility is becoming a central policy issue. U.S. regulators are signalling resistance to the idea that an AI agent can be treated as an independent actor when its conduct causes harm.

That approach would put pressure on developers and deployers to build stronger controls, monitoring and audit trails. For businesses, the age of “the AI did it” is unlikely to be an adequate governance strategy.