By 2026, AI isn't just assisting workers, it's acting for them. Autonomous agents now send messages, move money, write code, and make decisions at machine speed. But as AI becomes the fabric of how we work, a hard question surfaces: when an agent acts, who is accountable for what it does?
This session confronts the accountability gap emerging in the agentic era. When software takes actions on an employee's behalf, the lines around responsibility, privacy, and compliance blur, and most organizations are deploying these systems faster than they can govern them. That gap is where real-world risk lives: unauthorized actions, data exposure, and regulatory exposure that surfaces only after something goes wrong.
Drawing on frameworks like the NIST AI Risk Management Framework and the EU AI Act, Kevin walks through practical accountability models for autonomous AI, how to set guardrails, assign ownership, and build human oversight into agent workflows before regulation forces the issue. The focus is on what works in practice, not theory.

Key takeaways:

- Accountability models for deciding who owns the risk when AI agents act
- Emerging regulatory signals every leader should track in 2026
- A practical checklist for deploying autonomous AI safely and responsibly