GenAI moves into claims
Generative AI is no longer experimental in claims — firms like Verisk and Aviva are using GenAI to speed document review, summarize medical records, and halve review times on complex claims, turning AI into a productivity and adjudication play rather than just a lab project. Regulators and underwriters are watching closely, so explainability and audit trails are being table stakes. ( )
Verisk introduced a generative-AI record-summary capability in Discovery Navigator on Feb. 8, 2024, claiming auto-summaries that can be up to 90% faster than manual review and achieve up to 95% accuracy. (verisk.com) Discovery Navigator’s auto-summary is built on Amazon Bedrock to call foundation models via API, and Verisk says the feature automates extraction and organization of key medical-treatment data for bodily-injury claims. (verisk.com) Verisk also launched XactAI as part of its Xactware suite on Sept. 30, 2025, packaging note summarization, automatic photo labeling and receipt categorization for property claims within workflow tools. (verisk.com) XactAI outputs are explicitly routed through human oversight and Verisk states the product is built with enterprise-grade security, encrypted processing and audit-trail support to meet insurer compliance needs. (verisk.com) Aviva announced on March 18, 2026 that it has extended its industry-first AI underwriting summarisation tool to individual critical-illness applications after a life-insurance rollout in November 2025, with chief underwriting officer Robert Morrison describing the CI extension as validated across a wider range of conditions. (aviva.com) Aviva has said its prior AI work in motor claims delivered more than £60 million in savings in 2024 and cut the average time to assess liability on complex motor cases by about 23 days, illustrating the firm’s quantified operational returns from automation. (mortgagesolutions.co.uk) The NAIC adopted a Model Bulletin on the Use of Artificial Intelligence Systems by Insurers in December 2023, and as of June 2025 some 24 states had fully adopted the bulletin’s AIS-program expectations for governance, transparency and third‑party/vendor controls. (bakertilly.com) Regulators in the UK (FCA/PRA) published AI‑strategy updates on April 22, 2024 mapping to government AI principles, and industry guidance plus NAIC exam activity now expects documented AIS programs, bias‑testing, vendor diligence and audit readiness. (fca.org.uk) Verisk’s public ethics and governance materials call out bias detection, data audits and clear accountability as design principles for AI-driven claims tools, underscoring the vendor-side emphasis on explainability and examinable documentation. (verisk.com)
