ML anti-fraud models and AI forgery detection (Gjensidige)
by Gjensidige Forsikring
Market unit: NO, DK
Gjensidige uses machine-learning anti-fraud models in claims in Norway as part of an AI-supported fraud prevention operating model, which produced NOK 71 million in savings; compared with 2022 the number of controlled cases rose by 65% and the hit rate by 40% by 2024. An AI-based forgery detection tool, used in lead handling and by investigators, exposed an AI-generated fraudulent claim document for the first time. Rollout of the fraud models to Denmark is planned; in Denmark an AI tool for recourse assessment raised recourse claims by 9% in 2024.
1 free view left this month
AI capability and maturity
AI capability level
Business context
- Value chain steps
- Claims managementFraud detection and prevention
- Insurance branches
- Property insuranceMotor vehicle insurance
- Application scope
- Internal only
- Country or market unit
- NODK
Value and impact
- Added value category
- Cost reductionRisk reduction
- Added value estimation
NOK 71m savings in Norway
- Quality gain
Controlled cases +65%, hit rate +40% vs 2022
Technical details
- AI category
- Machine learning and predictive analyticsAnomaly detection
Compliance and governance
- EU AI Act risk category
- Minimal risk
- EU AI Act role
- Deployer
- Relevant policies
- EU Artificial Intelligence ActGeneral Data Protection Regulation
Metadata
- Development type
- Scaled application
Community
Average rating
Comments
Log in to join the discussion.
No comments yet. Be the first to comment.
Related solutions
FraudOps is a UK-based fraud case management and investigation tool, backed by Toolagen Venture Studios, that ingests alerts from detection systems such as FRISS, triages referrals and manages the full case lifecycle with SLA tracking. Generative-AI agents assemble referrals, run intelligence and watchlist searches, and draft investigation notes and reports; the vendor states it does not build predictive models. The vendor's website says it is live with Zurich Insurance and quotes a Zurich Claims Fraud Operations Manager. No Zurich, trade-press or award confirmation was found and the Zurich unit is not named, so this is a vendor claim only.
Zurich Chile uses SmartClaims LISA, a generative-AI and no-code decisioning platform from Chilean insurtech LISA, to automate health-claims reimbursement end to end. Zurich Chile handles about 350,000 health claims a month; by end-August 2024, 33% of traditionally reimbursed claims (roughly 33,000 a month) were processed automatically and settlement fell from about 4.5 days to within a day (Zurich, October 2024). Zurich's 2026 Qorus Innovation in Insurance Awards submission reports 65% end-to-end automation, settlement in 10 minutes, 35% cost reduction and +30 NPS points. More than 1.2 million claims were processed in 2025. The project was recognised at the Zurich Innovation Championship 2025 for impact in Latin America; the collaboration began through ZIC 2022.
Admiral Seguros, the Spanish unit of the UK-based Admiral Group, uses Tractable's computer vision for straight-through motor claims. Customers photograph the vehicle damage via a web-app link; the AI assesses the damage and builds a repair estimate, so a claim can move from notification to settlement offer without an appraiser. Tractable reports 12,000 touchless claims in 2021, 90% of estimates processed without appraiser input, 96% accuracy compared with human appraisers and 98% of qualifying claims completed in under 15 minutes; 70-75% of customers who receive the link complete the claim.