Some Insurers Don’t Want to Cover AI’s Errors

Some in the commercial insurance industry don’t want to cover losses caused by artificial intelligence chatbots. (Dreamstime)

Key Points

• Insurers are introducing exclusions for artificial intelligence in new policies, fearing liability from generative AI errors.
• Verisk Analytics plans to offer its insurance clients forms in 2026 to exclude generative AI-related bodily injury, property damage, and personal injury.
• Some insurers, like AIG, say they are not currently planning to implement gen-AI exclusions.

As businesses hustle to incorporate artificial intelligence in everything they do, their insurers want no part of it. Across the commercial insurance industry, new policies say they won’t cover losses caused by chatbots.

Early this year, word spread that the specialty insurer W.R. Berkley was putting an “absolute” exclusion of coverage for “any actual or alleged use, deployment, or development of Artificial Intelligence” in some liability policies for corporate directors and officers.

Other insurers may follow, for fear that erroneous price quotes or medical diagnoses from a generative AI model could lead to thousands of claims.

AI exclusions could become an industry standard next year. In 2026, the consulting firm Verisk Analytics will provide its insurance customers with forms that can generally exclude this “emerging exposure”.

Verisk’s forms would exclude coverage for bodily injury, property damage, or personal and advertising injury caused by generative AI. It defines gen-AI as “a machine-based learning system or model that is trained on data with the ability to create content or responses, including but not limited to text, images, audio, video or code.”

If that sounds like every chatbot offered by OpenAI, Meta Platforms, Alphabet’s Google, or Anthropic, that’s clearly what the insurers intend.

A Sunday article in the Financial Times reported that standardized forms with similar gen-AI exclusions had been filed with state regulators by a local unit of American International Group and the Great American Insurance unit of the American Financial Group.

But AIG tells Barron’s that the forms’ gen-AI provisions didn’t reflect the insurer’s intent. “AIG was not specifically seeking to use these exclusions and has no plans to implement them at this time,” said an AIG spokesperson.

Whether the insurance industry will succeed in blocking all AI claims is far from settled. Just what is an “AI use”, and what is “AI”, in the first place, remain subject to interpretation, said a May 2025 note by lawyers at the national insurance litigation firm Hunton Andrews Kurth.

“Insurance policies are sold by insurance brokers,” wrote the lawyers. “They’re bought by risk managers; claims are handled by claim handlers and disputes are typically decided by judges.”

Write to Bill Alpert at william.alpert@barrons.com