What liability does my company have for decisions made by AI agents?

Answer

Under current law in most jurisdictions, your company bears full liability for agent actions — there is no 'the AI did it' defense. Key precedents: the Air Canada chatbot ruling (company liable for chatbot's incorrect refund policy), FTC enforcement actions against algorithmic bias, and emerging EU AI Act obligations for deployers. The liability chain typically flows: deploying organization then development team then model provider, but the deploying organization is almost always the primary target. Governance creates the evidence trail that demonstrates reasonable care.