AI and Treasury Management: More Opportunity, More Risk
Artificial intelligence is quickly finding its way into the tools businesses use every day. For finance teams, that creates real opportunities to automate routine work, make better use of financial data and eventually make payments more efficient.
But there’s another side to the story. Fraudsters have access to AI, too.
That doesn’t mean businesses should avoid AI. It means they should understand where it can help, where it can create risk and where human judgment still matters.
AI can take routine work off your team’s plate.
One of AI’s clearest benefits is its ability to reduce repetitive, manual work.
For treasury teams, that could mean helping create payment files rather than keying information manually or preparing data in the format needed to connect business systems with banking services. As these capabilities mature, AI could also play a larger role in intelligent payment routing and identifying or correcting errors before transactions are sent.
Reducing manual work is a boost for efficiency, giving employees more time for work that requires experience and judgment, and it can reduce opportunities for human error.
The same is true inside banks. As AI takes on more routine tasks, bankers can spend more time helping clients solve problems, evaluate options and plan for what’s ahead.
AI can make financial data easier to use.
Businesses increasingly want more than a traditional report. They want to ask questions of their financial data, quickly analyze the results and turn that information into graphs, visuals or presentations.
AI is making that possible.
Your bank doesn’t necessarily have to provide all those capabilities. A business’s enterprise resource planning, or ERP, system, accounting software or other financial technology may be better suited to analyze the data the bank provides.
Before adding a new AI tool, find out what your existing platforms can already do. And when evaluating an AI-enabled financial tool, ask questions such as:
- How is our information stored and protected?
- Is our data used to train an AI model?
- Who can access our information?
- Is data encrypted and what security or audit standards does the provider follow?
- What happens to our information if we stop using the service?
Most importantly, don’t put confidential company, customer or financial information into an AI tool that your business hasn’t approved.
Fraudsters get an efficiency boost, too.
The qualities that make AI useful to businesses also make it useful to criminals. Tasks that once took considerable time can now be performed quickly and at enormous scale.
Consider business email compromise, or BEC. AI can help criminals write more convincing emails, personalize messages using information found online and imitate the writing style of an executive or vendor. Voice cloning can make an impersonation attempt even more believable.
The technology has changed, but many of the best defenses have not.
- Be skeptical of unexpected requests involving payments or account changes, particularly when someone creates a sense of urgency.
- Verify changes using a known, trusted phone number rather than contact information supplied in the request.
- Maintain strong payment controls and make sure employees know when and how to stop a transaction for additional verification.
AI can make a fake request more convincing. It shouldn’t make your business more willing to bypass its controls.
Protect the information you give AI.
Fraud isn’t the only AI-related risk. Businesses also need to consider what employees may unintentionally share with AI tools.
Putting non-public information into an unapproved AI service can move sensitive data outside the systems your company has established to protect it. That could include financial information, customer data, internal documents or other confidential material.
Businesses should establish clear rules about which AI tools employees may use and what information can be shared with them. For financial applications in particular, understand a provider’s security practices before trusting it with sensitive data.
Keep people in the process.
AI will become more capable, but there are decisions treasury clients should never delegate entirely to technology.
Authorization, verification and judgment still require people and strong controls. AI can help prepare information, identify patterns and reduce manual work. A person should still be accountable for sensitive financial decisions and unusual requests.
That’s a useful way to approach AI overall: Use it to reduce manual work, not human judgment.
Businesses don’t have to choose between innovation and security. A thoughtful approach can provide both.
A good treasury advisor can help you review your current processes, identify opportunities to improve efficiency and make sure the right controls remain in place as the technology evolves.
Greg Skrzypek is a treasury management advisor based at Pinnacle’s Riverwood office in Atlanta. He can be reached by phone at 470-936-7569 and by email at Greg.Skrzypek@pnfp.com.
