AI is moving quickly into everyday business operations, but the insurance industry is starting to draw a harder line around what it is willing to cover. According to reporting from the Financial Times, major insurers are asking U.S. regulators for permission to exclude broad AI-related liabilities from corporate policies.
The concern is simple to state and difficult to price: when AI systems generate bad outputs, the consequences may not stay contained inside one company. They can spread across many users, policies and claims at the same time.
Insurers want room to limit AI liability
Major insurers including Great American, Chubb, and W. R. Berkley are seeking permission from U.S. regulators to exclude widespread AI-related liabilities from corporate policies. The move reflects a growing discomfort with how unpredictable AI failures can be, especially when the same tools are used across many businesses.
One underwriter described the outputs of AI models to the FT as "too much of a black box." That phrase captures the practical problem for insurers: they are being asked to cover risks that may be difficult to understand before a claim arrives.
AIG was also listed in the FT story, but later sent TechCrunch a statement clarifying its position: "AIG was not specifically seeking to use these [reported upon] exclusions and has no plans to implement them at this time."
Why AI risk is different from ordinary software risk
Companies have long used software to make decisions, serve customers and automate work. The issue raised by insurers is that AI outputs can be harder to predict, harder to explain and potentially easier to misuse at scale.
The source examples show several different kinds of exposure:
- False information: Google’s AI Overview falsely accused a solar company of legal troubles, triggering a $110 million lawsuit back in March.
- Invented customer promises: Air Canada last year got stuck honoring a discount its chatbot invented.
- AI-enabled fraud: Fraudsters last year used a digitally cloned version of a senior executive to steal $25 million from the London-based design engineering firm Arup during a video call that seemed entirely real.
These examples are not the same type of failure, but they point to the same larger problem. AI liability can come from inaccurate content, automated customer interactions or convincing synthetic media used by criminals.
The bigger fear is systemic risk
For insurers, the biggest concern is not necessarily one very large payout. The deeper worry is systemic risk: many claims happening at the same time because many companies rely on the same or similar AI systems.
As one Aon executive put it, insurers can handle a $400 million loss to one company. What they cannot handle is an agentic AI mishap that triggers 10,000 losses at once.
That distinction matters. A single corporate loss can be evaluated, priced and absorbed within traditional insurance thinking. A wave of simultaneous claims caused by a widely used AI model is much harder to plan for, because the exposure may be shared across many insured businesses at once.
What this means for companies adopting AI
The insurance response is a warning sign for companies that are rushing to adopt AI in customer service, search, decision support, internal workflows and automated agents. If insurers narrow coverage, businesses may carry more of the risk themselves when AI systems produce harmful results.
The source does not say that AI cannot be insured in every case. It says major insurers are asking regulators for permission to exclude widespread AI-related liabilities from corporate policies. That is a more specific, but still important, development.
For companies, the practical implication is that AI adoption cannot be treated only as a technology decision. It also creates legal, operational and financial exposure. If an AI system invents a discount, makes a false claim about another company or is used in a convincing fraud, the cost may not stop with a technical fix.
The unresolved question
The insurance industry exists to price risk. When insurers say a category of risk may be too uncertain or too broad to cover in standard corporate policies, it signals that the market has not yet settled on how to manage that exposure.
AI systems are being adopted because they can act quickly and at scale. The same qualities can make failures harder to contain. That is why insurers are focused less on one dramatic incident and more on the possibility of many businesses facing losses at once.
For now, the central tension is clear: companies want the productivity gains of AI, while insurers are questioning whether the downside can be priced in the usual way.