Major Financial Firms Revolutionize Fraud Detection and Credit Scoring with AI-Powered Transaction Models
In brief
- Financial institutions in 2026 have made significant strides in fraud detection and credit scoring by using large-scale transformer models trained on billions of transaction sequences.
- Companies like NVIDIA, Stripe, Nubank, Visa, Mastercard, Revolut, and Plaid have developed tools that enable these advancements, with NVIDIA's Build Your Own Transaction Model leading the way.
- This model uses GPU acceleration and custom tokenization to preprocess data, then trains a compact Llama-based decoder-only AI system.
- A near-50% improvement in accuracy over traditional methods on IBM's TabFormer fraud dataset.
- These models are transforming how financial tasks are handled.
- Instead of relying on outdated rule sets and hand-engineered features, foundation models analyze sequential customer behavior to create robust representations for various applications-like fraud detection, credit scoring, and personalized recommendations.
- The shift is accelerating across the industry, with firms reporting double-digit performance gains while reducing operational complexity.
- Looking ahead, expect more financial institutions to adopt these AI-driven approaches, expanding their use in areas like customer segmentation and transaction pattern analysis.
- The integration of raw data features with pre-trained embeddings promises even greater efficiency and accuracy in fraud detection and beyond.
Terms in this brief
- GPU acceleration
- Using graphics processing units (GPUs) to speed up computing tasks, especially in AI and machine learning where parallel processing is crucial for efficiency and performance.
- Custom tokenization
- A method of converting raw data into tokens that a model can understand. In this context, it's tailored to better process transaction sequences for fraud detection and credit scoring.
Read full story at NVIDIA Dev Blog →
More briefs
California Gas Stations Accused of Using AI to Raise Prices
A lawsuit claims many California gas stations used an AI tool to inflate prices. The lawsuit says this caused prices to surge in the state. The suit claims the AI tool allowed gas stations to avoid competing with other stations and charge higher prices to consumers. This could make gas prices rise by as much as 30 cents per gallon in some areas. The average gas price in California is $5.52 per gallon, the highest in the country. The lawsuit was brought against 1,732 gas stations in the state, including those owned by Walmart and Albertsons. The stations are accused of using the AI tool to set prices, which is illegal under a 2025 law. Gas prices in California will likely be watched closely as the lawsuit moves forward.
AI Beats Lawyers in UK Court
A woman used artificial intelligence to win a court case in the UK. She paid about $529 for the AI help and won $9,271 in unpaid fees. The AI handled pretrial work and a human argued the case. This is the first trial won by an AI law firm in the world. The AI law firm has processed over 600 claims and recovered about $662,000 for clients. The AI made the process more accessible and affordable. The woman was delighted by the result and said she could pursue the claim thanks to the AI. The use of AI in law will continue to grow.
NVIDIA Introduces Energy Efficient AI Servers
NVIDIA has introduced new AI servers that can run their cooling liquid at up to 45 degrees Celsius. This higher temperature limit makes them more energy efficient. The new servers are part of the Rubin generation of NVIDIA AI infrastructure. The new servers are the world's first to achieve 100% liquid cooling. This means every chip and component is cooled by liquid in a closed loop with no fans. Historically, cooling has accounted for up to 40% of a data center's electricity consumption. Raising chiller plant temperatures by just one degree can cut cooling energy costs by about 4%. A 50-megawatt hyperscale facility can save over $4 million annually in cooling-related energy and water costs. The new liquid-cooled infrastructure can reduce facility cooling water consumption from roughly 2.6 million gallons per megawatt per year to near zero. NVIDIA's 45-degree liquid cooling will enable data centers to operate more efficiently in the future.
Reflection AI Partners with SpaceX for AI Chips
Reflection AI will pay $150 million a month for access to Nvidia's latest AI chips. The deal is worth up to $6.3 billion and will last through 2029. The deal matters because it shows the value of open source AI. Reflection AI used this deal to promote its open-weight AI strategy. This strategy is an alternative to closed models like those used by Anthropic and OpenAI. The company will use the AI chips to build open models at scale. Reflection AI will have more computing power to work on its projects. The company will start using the AI chips on July 1, 2026.
NVIDIA Launches Safety System for Robots
NVIDIA has launched a safety system for robots called Halos for Robotics. This system is the first full-stack safety system for robots and physical AI. The system is important because it helps robots work safely with humans. Over 18,600 engineering years of autonomous vehicle safety development went into creating this system. NVIDIA will continue to work with companies to build safer robots using this system.