Visa boosts security with AI-powered data retrieval

Visa’s enterprise-level deployment of generative AI showcases how financial giants can implement cutting-edge technology while maintaining strict security protocols. By building a protected, multi-model AI environment behind its firewall, Visa has accelerated critical business processes while addressing the natural tensions between innovation, security, and regulatory compliance in the financial sector.

The big picture: Visa has developed a secure generative AI system that operates within its company firewall, allowing employees to use AI tools while protecting sensitive financial data.

• The company’s “Secure ChatGPT” runs internally on Microsoft Azure, featuring controlled input and output screening to prevent data leakage.
• This implementation addresses the overwhelming employee demand for AI tools while maintaining the strict security requirements essential for a global payments processor.

Why this matters: By implementing RAG (retrieval-augmented generation) technology, Visa has dramatically accelerated access to critical regulatory information across its 200+ global markets.

• What previously took days of manual research now happens up to 1,000 times faster, allowing client services teams to handle significantly more cases.
• This efficiency gain directly impacts business operations while ensuring accuracy through source citation.

Key details: Visa’s internal AI system offers employees access to six different AI models while maintaining complete control over data security.

• Available models include GPT, Mistral, Claude, Llama, Gemini, and Granite, all running behind Visa’s firewall.
• The system incorporates data loss prevention (DLP) screening to ensure no sensitive customer information leaves Visa’s protected environment.

Behind the numbers: Visa has invested over $10 billion in fraud prevention and network security technologies, with AI playing an increasingly central role.

• These investments helped the company block $40 billion in attempted fraud in 2024 alone.
• The company uses deep learning neural networks and synthetic data simulations to enhance its fraud detection capabilities.

What they’re saying: “First of all, it’s better quality results,” Sam Hamilton, Visa’s SVP of data and AI, told VentureBeat regarding their RAG implementation.

• Hamilton noted that when ChatGPT was released in November 2022, Visa employees immediately began asking: “Where is my ChatGPT?” and “Can I use ChatGPT?”
• The company recognized both the demand and the risk, developing their secure solution to balance innovation with their responsibility as a global payments provider.