Mastercard Adapts Fraud Detection Systems for AI-Driven Transactions

Evolving Fraud Detection for AI Buyers

Mastercard is updating its fraud detection systems to differentiate between malicious and legitimate bot-initiated transactions. Historically, these systems were designed to block all bot activity due to its association with fraud. However, with the rise of AI in legitimate purchasing, Mastercard needs to enable these transactions while maintaining security.

Greg Ulrich, Mastercard's chief AI and data officer, stated at VB Transform 2026 that the company's risk framework and rules require modification to support this new landscape. This represents a significant shift from the previous approach of preventing all bot transactions.

Transaction Scoring and AI Integration

Mastercard's network evaluates each transaction for fraud risk within 100 milliseconds, assigning a score from 0 to 999. This score is then passed to the issuing bank. The company processed 175 billion such transactions last year.

The integration of generative AI has enhanced this process, allowing Mastercard to incorporate more data and context into its risk assessments. This has led to a 300-400% increase in the identification of fraudulent transactions in high-risk categories without increasing friction or false positives for consumers.

Impact on Mastercard's Services Business

Beyond fraud detection, AI plays a crucial role in Mastercard's broader services business, which now accounts for approximately 40% of the company's operations. These services include marketing, fraud prevention, security, and business intelligence.

Currently, one-third of these services are predicated on AI. Mastercard is also developing its own transformer model based on its transaction data to build new solutions for safety, security, and personalization.