Mastercard spent decades training its fraud system to see bots as thieves. Now bots are the ones doing the buying.

Mastercard has spent decades perfecting its fraud detection system, which can analyze 175 billion transactions per year in under 100 milliseconds. However, with the rise of generative AI, the company is now faced with the challenge of enabling bots to make purchases while maintaining security and preventing fraud. Greg Ulrich, Mastercard's chief AI and data officer, explained that the company has had to modify its risk framework and rules to accommodate this new reality.

The company's Safety Net system has already stopped over 70 billion fraudulent transactions, and Mastercard is building its own transformer model on its transaction data to power new safety, security, and personalization solutions. The stakes extend beyond merely preventing fraud, as about 40% of Mastercard's business is now based on services, including AI-driven fraud and security solutions.

As agentic commerce grows, with more agents delegating authority on behalf of consumers and businesses, trust becomes increasingly crucial. Mastercard has built five layers of security to address these challenges: identity verification, verifiable intent, controls, execution through Agent Pay, and intelligence spanning risk rules, insights, and monitoring.

The ultimate goal is to enable agentic commerce, such as automated procurement agents managing inventory and supplier relationships. To achieve this, Mastercard must establish clear standards for identity, intent, and communication across various parties, ensuring a robust trust infrastructure for autonomous transactions.