Insurance and Artificial Intelligence: From Pricing Better to Preventing Loss, and the Rules That Will Shape the Shift

Purpose. Insurance sells predictions about risk, so it is a natural home for artificial intelligence, and a sensitive one, since predictions that sort people by risk can also sort them unfairly. This article asks where AI stands in insurance in 2026 and where the industry may be by about 2035. Design and methods. A review that combines the share of the insurance research literature involving AI (OpenAlex, 2010–2025), a supervisory survey of AI use in UK financial firms, the academic literature on AI along the insurance value chain, and the state of regulation in the European Union and the United States. Claims taken from the literature or from official documents were checked against the sources cited. The scenarios are judgements built on this evidence. Findings. The share of insurance works that mention AI rose from 1.1% in 2018 to 8.0% in 2025 (7.0 times as much), and language-model works from 9 to 382. In the 2024 Bank of England and FCA survey, insurance had the highest share of firms using AI of any financial sector, 95%. A review of 91 academic papers and 22 industry studies concluded that AI may move insurance from loss compensation towards loss prediction and prevention, may reduce information asymmetry, and may change the risk landscape itself. Regulation is taking shape unevenly: AI used for risk assessment and pricing in life and health insurance counts as high-risk under the EU AI Act, with obligations applying from 2 December 2027, and 29 US jurisdictions had adopted the NAIC model bulletin on insurers' use of AI by October 2026. We expect claims and underwriting automation to be routine by 2028, the debate about fairness and risk pooling to intensify between 2029 and 2032, and the largest long-term question to be how AI changes the risks that insurers cover. Originality. The article links a widely used adoption figure for insurers with the academic evidence on what AI changes in the business model, adds the current state of regulation in two jurisdictions with counts taken from official documents, and states clearly which parts of the outlook rest on evidence and which are judgement.