Mastercard launches transaction data-driven generative AI system to enhance security, insight and personalization
Alyssa Davidson
Posted: March 25, 2026 6:47 AM Updated: March 25, 2026 6:47 AM
Edit and fact check date: March 25, 2026, 6:47 AM
briefly
Mastercard is developing generative AI-based models trained on anonymized transaction data to improve insights, fraud detection, and payment services while protecting user privacy.
Mastercard, a technology company and global payments network, has introduced a generative AI system designed as a large-scale foundational model to support a wide range of applications. The model is being trained on a proprietary dataset derived from billions of payment transactions, with personal identifiers removed to protect user privacy. By analyzing anonymized patterns within this data, the system is designed to generate insights and predict future trading behavior.
This approach is similar to modern conversational AI systems that predict a series of subsequent words, but in this case the model is not intended for conversation generation. Instead, it is being developed as an analytics engine to enhance existing services, including cybersecurity measures, customer loyalty programs, and tools for small and medium-sized businesses.
The system is being developed with support from leading compute and data infrastructure providers, including Nvidia and Databricks, enabling large-scale processing and accelerated model training. The company said it expects the results of this work to be presented at an upcoming industry conference.
Foundational AI models built on structured transaction data to enhance payments and security
The underlying architecture differs from commonly used large-scale language models that are trained on unstructured data such as text, images, and video. Instead, these models fall into a category called large-scale tabular models, which are trained on structured data sets organized into tables. The training process incorporates large-scale transaction data, and we plan to expand to broader data sets, including merchant location information, fraud indicators, authorization history, chargeback data, and loyalty program activity.
Increasing data coverage improves the model’s ability to identify patterns and produce more accurate predictions. One of the key areas of focus for application is cybersecurity. In cybersecurity, existing systems are already being used to detect and prevent fraud. The integration of this new model is expected to enhance these capabilities through improved pattern recognition and reduced false positives.
Current cybersecurity models typically rely on engineering features created by data scientists to highlight specific signals within transactional data, such as sudden changes in spending behavior. In contrast, new systems are designed to learn these patterns with minimal manual feature engineering, allowing them to identify relationships in the data that may not be immediately obvious with traditional methods.
Initial testing has shown improved performance over traditional machine learning approaches in reducing false positives, especially in scenarios involving legitimate but uncommon transactions. The system has proven its ability to better distinguish between unusual but valid activity and potentially fraudulent activity.
Additional potential applications include improving personalization systems, optimizing rewards programs, improving portfolio analytics, and advanced data analytics capabilities. This model is also expected to reduce the need to maintain large numbers of specialized models across different geographies and use cases.
Future development plans include expanding model functionality, improving architecture, and introducing application programming interfaces and developer tools for broader use across the organization. Ongoing collaboration with technology partners is expected to support continued advancements.
The plan is being developed in line with established data governance principles, emphasizing privacy, responsible data use, and transparency. As development progresses, this model is expected to contribute to greater efficiency, greater security, and greater intelligence within payment and commerce systems.
disclaimer
In accordance with the Trust Project Guidelines, the information provided on these pages is not intended and should not be construed as legal, tax, investment, financial or any other form of advice. It is important to invest only what you can afford to lose and, when in doubt, seek independent financial advice. Please refer to the Terms of Use as well as the help and support pages provided by the publisher or advertiser for more information. Although MetaversePost is committed to accurate and unbiased reporting, market conditions may change without notice.
About the author
As MPost’s resident journalist, Alisa specializes in the broad areas of cryptocurrencies, zero-knowledge proofs, investing, and Web3. With a keen eye for new trends and technologies, she provides comprehensive coverage to inform and engage readers about the ever-evolving digital financial landscape.
more articles
As MPost’s resident journalist, Alisa specializes in the broad areas of cryptocurrencies, zero-knowledge proofs, investing, and Web3. With a keen eye for new trends and technologies, she provides comprehensive coverage to inform and engage readers about the ever-evolving digital financial landscape.
more articles