When the AI Acts, Not Just Advises

What this session covers

• The point in a deployment where conversational AI moves from advising to acting, and how the liability profile shifts the moment it does
• The control architecture senior leaders are now building around action-taking agents, including authorisation, confirmation, audit trails, and reversal pathways
• The cross-jurisdictional regulatory pressure on autonomous action across the EU AI Act, US state-level AI law, and emerging APAC frameworks
• How the most mature programmes are deciding which actions an AI can take unilaterally, which require human confirmation, and which the AI should never own