Insurers face an AI paradox

Insurance depends on predictable, independent risks. Agentic AI is unpredictable and shared across thousands of companies simultaneously, making traditional risk pooling nearly impossible.

One AI failure, thousands of claims

When thousands of businesses rely on the same foundation model or cloud infrastructure, a single defect upstream could trigger correlated losses across supposedly independent policyholders at once.

CFOs are already deploying agents

Nearly 7% of US enterprise CFOs had deployed agentic AI in live finance workflows by September 2025, with another 5% running pilots, according to PYMNTS Intelligence.

Multiple liability types emerge

A malfunctioning agent could trigger professional liability claims, cyber losses, or directors-and-officers issues depending on the failure mode and how management oversaw deployment.

Insurers become AI gatekeepers

Companies seeking coverage will need to prove how models are tested, what systems agents can access, whether humans approve major actions, and whether incidents can be audited afterward.

“An enterprise may technically be able to let an agent autonomously move $10 million, modify production code or negotiate contracts. Its insurer may simply decline to cover that configuration without additional controls.”

Coverage conditions set autonomy limits

The practical ceiling on AI autonomy may be set not by what technology can do, but by what risk markets will finance through insurance coverage and premium requirements.

Tech vendors vs. buyer responsibility

Technology companies compete on capability and autonomy. Corporate buyers must compete on how much autonomy they can responsibly put into production while staying insurable.

Existing laws still apply to AI

Older legal obligations around liability, compliance and governance still apply to AI systems, creating additional risk layers that insurers must evaluate and price.