SwedenFinance and insurance2026 – ongoingTrustly
Bank activity categorization for payment fraud detection
Trustly reduced fraud detection false positives by 43% using generative AI
Problem
Trustly needed to distinguish legitimate payments from fraud across large amounts of banking data. Its detection models needed better information about account activity.
Approach
Trustly built its own fraud detection models with Amazon SageMaker. It added Amazon Bedrock to categorize bank activity for those models through its Trustly Insight feature. The system operated in its US payments business.
Results
Takeaway
Trustly improved the information feeding its existing fraud models, reducing false positives in payment decisions.
Categories
- Industry
- Finance and insurance
- Business function
- Finance, risk and legal
- Innovation type
- Process innovation
- AI technology
- Generative AIPredictive analytics and ML
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
AWS described generative AI categorization supporting Trustly's fraud models. It attributed a 43% reduction in false positives to generative AI identification of emerging fraud patterns. The page carries no date; the Internet Archive first saved it on August 25, 2026.
Full details are in the original sources.
