Affirmative AI insurers expand their offerings
(The Insurer) - The first companies to launch standalone insurance products for artificial intelligence risks have already begun to expand their product offerings as the nascent market develops.
It is not yet clear whether growing AI liability risks will be addressed by existing product categories or by a completely new class of insurance. Even if AI does follow in the footsteps of cyber to be underwritten separately to other classes, it will likely take time for a market to form around standardised products.
The industry has seen this pattern with carbon credit insurance in recent years, as insurers and brokers have tested products covering varying stages of the carbon credit lifecycle for different organisations in the carbon market.
Only now, according to CFC's head of innovation George Beattie, is carbon credit insurance beginning to move past the "rapid prototyping" phase and approaching the "product-market fit proof point".
Currently, AI insurance is in the rapid prototyping phase.
Munich Re became one of the first (re)insurers to offer AI-specific coverage when it underwrote a performance guarantee for a fraud prevention AI model in 2018. It soon expanded this principle to guaranteeing the performance of other types of commercial AI models.
By 2023, the reinsurer was offering underperformance insurance to companies implementing AI models they had developed in-house.
The same year, Armilla AI, a company then focused on AI risk management and evaluating performance, entered the tenth cohort of the Lloyd's Lab accelerator programme.
It subsequently launched insurance-backed performance guarantees for AI product providers in late 2023, backed by capacity from Swiss Re, Greenlight Re and Chaucer.
By 2025, both Munich Re and Armilla were also providing AI insurance for enterprises' usage of AI models, covering various financial and liability risks associated with deploying the technology.
Testudo, the latest AI insurance startup to graduate from the Lloyd's Lab, plans to launch coverage for enterprises deploying AI too.
Its co-founder and CEO George Lewin-Smith told The Insurer that the reasons it focused on that client segment included that those companies were struggling the most with the risks of deploying AI, and that it would help control accumulation of risk as an underwriter, compared to insuring a model or vendor.
The insurance market currently remains in the "investigative phase" of determining how to handle exposures to AI, in the words of Armilla co-founder and CEO Karthik Ramakrishnan in a recent interview with The Insurer TV.
If AI insurance becomes a product category in its own right, it will take time to establish both the scope of AI-related risks that should be covered and in what circumstances the coverage would be purchased by technology providers or those deploying AI.
This may take some time, with the current soft cyber market acting as a brake on insurers' ability to exclude AI risks and move them into a new coverage category, lawyer Howard Panensky from Pierson Ferdinand said.
The Insurer's discussions with Testudo's Lewin-Smith, Armilla's Ramakrishnan and Pierson Ferdinand's Panensky are part of a series of articles published this week focused on emerging AI risks.
