As businesses and individuals increasingly use AI for advice and guidance, it is clear that these tools cannot always be trusted. For example, September 2026 research from AI specialist Saturn found that over half of the financial advice provided by 18 AI tools it tested was wrong. This situation could lead to both opportunities and threats for insurers.
GlobalData’s 2026 UK SME Insurance Survey found that among SMEs that are interested in an AI liability policy, 47% would want cover for legal fees resulting from AI-related decisions as part of the policy. This was the most popular policy feature in the survey, ahead of coverage for reputational damage and regulatory fines.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
The study undertaken by Saturn found that wrong advice was given for 53% of the 121 financial-related questions asked to 18 AI tools (including Gemini, ChatGPT, and Claude). These results were for basic financial questions; for more complicated questions, correct advice was provided just 12% of the time.
The fact that people and businesses are increasingly using these AI tools for financial advice poses a significant issue. Where liability falls for bad AI outcomes remains a grey area, but it is a space where insurers could step in and offer AI liability policies with incorrect advice coverage to businesses using these tools.
However, Saturn’s study suggests this would be a risky move for insurers at present. AI tools are not yet ready to be giving out unchecked advice in important areas such as finance. In addition, there is little historical data to rely on, which makes pricing difficult.
GlobalData surveying shows that AI liability would be a popular insurance product for UK SMEs, while increased adoption of AI suggests it is a high growth opportunity for insurers. However, providers would need to understand the level of risk they are insuring against and confirm that checks are in place (such as human reviews of AI advice) at customers using these AI tools.
