Silent AI

Silent AI is the industry term for the coverage gap that exists when an insurance policy is neither affirmatively granting coverage for artificial intelligence claims nor explicitly excluding them. The policy is silent on AI, which leaves both the insurer and the insured uncertain about whether a generative AI loss falls inside or outside the form. The phrase mirrors the older silent cyber problem, in which traditional property and casualty policies were silent on cyber-attack losses and litigation followed.

The concept matters because most pre-2025 Commercial General Liability, Tech E&O, Cyber, and Professional Liability forms predate generative AI as a named exposure. Their wordings were drafted for risks the underwriter understood, and AI was not on the schedule. When a claim arrives that originates from a hallucinated AI output, an autonomous agent's action, or AI-generated infringing content, the silent policy becomes the battleground: the insured argues coverage; the insurer argues the loss was never contemplated.

Insurers have responded with two opposite moves. Some carriers explicitly affirm AI coverage in their wordings (often through narrow endorsements with sub-limits), converting silent AI into affirmative AI coverage. Others explicitly exclude AI claims through endorsements like Verisk's Generative AI exclusions (CG 40 47 and CG 40 48) attaching at U.S. renewals from January 1, 2026, converting silent AI into a confirmed exclusion.

For deployers of generative AI, silent AI is a poor position. The uncertainty is itself a risk: defense costs accrue while coverage is litigated, the carrier may reserve rights, and the outcome depends on the specific wording and the jurisdiction. The structural answer is to map every existing policy for affirmative or excluded AI language, and place a standalone Generative AI Liability policy where the underlying forms leave a gap.