How Is Wells Fargo Using AI Against Fraud?

Wells Fargo is deploying AI across its payments business on two fronts simultaneously, using it to generate intelligence from payments data while defending against the AI-powered fraud that the same technology has made significantly more sophisticated.

Ather Williams III, Head of Global Payments and Liquidity and Wholesale Digital at Wells Fargo, outlined the bank's AI strategy during the keynote address at American Banker's Payments Forum in San Francisco on May 5, 2026.

His framing of the AI platform's role was specific: intelligence on top of transactions. "Intelligence on top of transactions converts data on balances, accounts receivable, accounts payable, foreign exchange exposure, liquidity positions into cash flow forecasting, liquidity risk detection, pricing insights and growth opportunities," Williams said.

The platform is deliberately built to avoid vendor lock-in. Wells Fargo's AI infrastructure is model and cloud agnostic, meaning the bank can adopt newer, larger, or more specialized models as they emerge without needing to rebuild the underlying architecture.

That design decision reflects a recognition that the AI model landscape is moving faster than any single vendor relationship can accommodate.

What the Numbers Show

Williams disclosed specific metrics on the bank's AI deployment progress at the Payments Forum. Wells Fargo's AI platforms have enabled 85% of its use cases today, across 335 experiments conducted internally.

Of those, 26 are live in production, a conversion rate that reflects the gap between what banks are testing and what they are actually running at scale, a gap Williams said the bank is actively closing.

American Banker's 2026 AI Talent Shift survey, conducted among 206 banking professionals in March 2026, found that 66% of respondents classify AI as a high-strategic priority, context that positions Wells Fargo's 335-experiment deployment program as among the most advanced in the industry.

The bank has also deployed AI in its sales engagement function. Rather than simply capturing call data in the CRM, Wells Fargo has built a tool that goes beyond transcription, producing client insights, advisory recommendations, and specific guidance for bankers before and after client interactions.

"Having insights into the clients, of being advisory, providing recommendations, that's the path of travel," Williams said. On scaling AI governance, Williams was direct about what responsible deployment actually requires at an institution of Wells Fargo's size.

"We allow people to experiment, build, test and deploy, while maintaining security governance and oversight at scale. We all have our experiments. Everyone has Copilot at their desk. Everyone messes around with Claude. We need to be able to do it at scale."

The Google Agentspace Deployment

In August 2025, Wells Fargo announced an expansion of its strategic collaboration with Google Cloud, deploying AI agents across its entire workforce through Google Agentspace.

The move made Wells Fargo one of the first major commercial banks to comprehensively adopt AI agents across its operations, at a time when a PYMNTS Intelligence survey found only 15% of CFOs were considering agentic AI deployment, citing a lack of trust in ceding control to autonomous systems. The deployment covers four specific areas.

In corporate and investment banking, AI agents answer, triage, and summarize complex foreign exchange queries post-transaction, navigate policies and procedures across internal data sources, and provide real-time market insights to bankers and executives.

In contract management, Wells Fargo maintains approximately 250,000 documents related to vendor agreements. An AI agent can query across those documents to find contracts with specific clauses, payment terms, contract types, and other information that previously required manual review.

In customer service across digital, branches, and call centers, AI agents handle routine requests like balance inquiries and debit card replacements, allowing bankers to focus on more complex tasks and deeper client relationships. The agents can also provide tailored product recommendations to customers.

In internal operations, employees can use Google's NotebookLM through Agentspace to upload documents, presentations, and spreadsheets and query the AI agent for research and analysis, replacing keyword-based search with natural language queries across enterprise data.

AI is also reshaping the threat environment Wells Fargo's customers and commercial clients operate in, and the bank's fraud education and awareness function is responding accordingly.

A survey cited by Wells Fargo's Fraud Education and Awareness Program found that 79% of organizations reported being victims of attempted or actual payments fraud in 2024, up from 65% in 2022.

Generative AI is a significant driver of that increase, enabling higher quality phishing emails, more convincing deepfake calls and videos, and automated fraud attempts that can target more companies more frequently.

The Hong Kong deepfake case, where fraudsters extracted more than $25 million from a company after an employee authorized payments to criminals impersonating coworkers on a corporate video call, including a convincing deepfake of the company's CFO, shows how far the technology has advanced.

Wells Fargo's fraud education framework, authored by Fraud Education and Awareness Program Lead Anil G. Khilnani, recommends AI-based secure email gateways as an industry best practice, dual custody processes for payments and system administration, and robust payment verification workflows that require separate communication channels to confirm payment instruction changes.

Where Williams Sees the Payments Landscape Heading

Williams closed his Payments Forum keynote with a forward-looking statement that framed Wells Fargo's AI investment as preparation for a payments infrastructure that does not yet fully exist.

"New capabilities such as tokenized deposits, programmable liquidity and near-real-time or atomic settlement didn't exist five years ago, but they will be table stakes five years from now," he said.

The intelligence layer Wells Fargo is building through AI is designed to operate across that future payments infrastructure, converting the data those capabilities generate into actionable insights for treasury, liquidity, risk, and commercial banking functions.

The 26 production use cases live today are, by Williams' own framing, only the beginning of that build.