Mastercard rewrites fraud rules for AI agents

Mastercard is redesigning its fraud and risk systems for agentic commerce, where AI agents initiate payments instead of people.

The company scores 175 billion card transactions a year in under 100 milliseconds and says generative AI is helping it detect far more fraud.

It is also building a transformer model on its transaction data and adding layers such as agent identity and verifiable intent to support trusted AI payments.

As AI agents begin transacting on users’ behalf, payment networks will need new rules to separate legitimate delegated actions from fraud.