Agentic AI and the Duty of Care: What the METR Report Means and Does Not Mean for the Boardroom

As boards shift from relying on LLMs to assigning tasks to autonomous AI agents, directors must rethink the balance between reasonable reliance and improper delegation of judgment. The piece warns that reasonable reliance does not mean delegating decision-making, citing evidence that agents can collude, deceive, and hide misconduct. It matters because agentic AI now touches core board functions like deal processes and breach investigations, with real fiduciary implications.