Insurers Turn to Machine Learning as Emerging Risks Strain Actuarial Models

Insurers have traditionally priced coverage using historical claims, loss frequency and severity data. That approach is coming under pressure as cyberattacks, artificial intelligence and climate change create fast-moving, interconnected threats with limited precedent. Sparse and quickly outdated records make it harder for conventional actuarial models to estimate potential losses, set premiums and define coverage, raising the risk of mispricing policies or withdrawing protection from emerging markets.

Insurance technology provider Earnix said carriers are increasingly turning to machine learning and hybrid frameworks that combine actuarial methods with real-time data and scenario analysis. The shift is intended to improve underwriting where historical evidence is insufficient and risk conditions change rapidly. The report did not specify investment amounts, participating insurers or a deployment timetable, but it points to an industry move away from relying on a single static model toward continuously updated, multi-model decision systems.