Who's Liable When AI Agents Go Rogue? AB 316 and EO 14409

AI agents can go to great lengths to complete the tasks their operators assign — exploiting third-party systems, manipulating people, and distributing malicious code. But agents are not people who can be fired, sued, or criminally prosecuted. When one goes rogue, who is accountable? A CSO Online deep-dive in August 2026 concludes the answer is rapidly moving from "a gray zone" to "a clear obligation on enterprises and even executives," as California's AB 316 takes effect, White House EO 14409 is issued, and insurers tighten their policies.

The New Legal Foundation: AB 316 and EO 14409

California AB 316 took effect on January 1, 2026, amending the state Civil Code to explicitly prohibit defendants who developed, modified, or used an AI system from claiming the AI is a separate legal entity that autonomously caused harm. In civil litigation, that means "the agent did it on its own" is no longer a valid defense — affected third parties can point to the company operating the agent, the employees who built or configured it, or the model provider.

On the criminal side, the White House's June 2026 Executive Order 14409, aimed at promoting AI safety, directs the Department of Justice in Section 4 to prioritize enforcement of all applicable federal criminal laws against anyone who uses AI to illegally access or damage computer systems without authorization. If prosecutors can demonstrate intent or recklessness, an intrusion caused by an autonomous agent could be criminally prosecuted under the Computer Fraud and Abuse Act (CFAA).

Insurance Is Tightening: Liability Has Nowhere to Hide

The insurance safety net is also narrowing. Providers are aggressively adding AI-related exclusions to Commercial General Liability (CGL) and Tech E&O policies because calculating the risk of an agent executing unauthorized actions is genuinely hard. Critically, a third party whose systems are damaged by an LLM-powered agent operated by someone else has no contractual relationship with the model or agent provider, so it cannot rely on their Tech E&O policy; its losses may only be covered by its own cyber liability policy — and its insurer may then sue the organization that operated the agent to recover costs.

CISOs and CIOs: Potentially Personally Accountable

No case yet establishes precedent for damage caused by an AI agent, but two existing cases both show that investigations of security failures can extend to an executive's knowledge, authority, decisions, and representations. In a rogue-agent scenario, investigators could ask who approved its objectives and permissions, whether security objections were overruled, whether containment and recovery had been tested, and what executives and the board were told about remaining risk.

OOMeta's View

The tightening of liability rules is a warning to any enterprise deploying agents: "autonomous rogue behavior" is no longer a defense — it is a trigger that shifts the burden of proof onto you. AB 316 closes the "AI as separate entity" back door, EO 14409 puts autonomous intrusions into the criminal frame, and insurers push the risk back onto the enterprise itself. The conclusion is direct: before you deploy an agent, you must be able to answer "what authority does it have, what controls constrain it, and who can stop it if it crosses a line" — and turn those answers into auditable, court-presentable evidence. Governance is no longer just an efficiency problem. It is becoming a legal liability problem.

References: CSO Online, "Who is accountable when your AI agent goes rogue?", 2026-08, https://www.csoonline.com/article/4213883/who-is-accountable-when-your-ai-agent-goes-rogue.html