Bad actors are increasingly using AI to create fraudulent documents when applying for business loans, and lenders must implement multilayer checks and technology to root out rotten applications.

“AI has had a profound impact on just making things look a whole lot more real,” Patrick Lord, senior project manager at online financial services company Rapid Finance, told FinAi News.

Fraudsters are increasingly using AI to forge documents to resurrect or take over existing companies. They may create a “zombie business,” posing as a real company no longer in business, or perform a takeover, pretending to be a company still in operation, according to data company Enigma.

Among business entities 10 years old or older in the United States, 2.5%, or 140,000, are zombies, meaning they have valid registration but no operations and no revenue, according to a December 2025 report from Enigma.

AI has lowered the skill barrier for fraudsters and allows them to scale operations, submitting more false loan applications to more banks, Lord said. He estimates that such fraud occurs at least a handful of times each week, with varying degrees of success.

The financial industry does not track zombie business fraud as a separate category, often filing such forms of fraud as synthetic business fraud, business identity misuse, first-party fraud, document fraud or application fraud.

“Small-business financial fraud is up, digital fraud dominates, document forgery is becoming more AI-enabled and the Fed has specifically called out synthetic business fraud as a growing risk,” Lord said.

Warning signs

Despite fraudsters using AI to create convincing documents, they aren’t perfect. While the business will appear legitimate at first glance, Lord advises lenders to look for discrepancies such as:

• Gaps in history;
• New domain registrations; and
• Suspicious typos.

In cases of attempted fraud, “chances are, when you go through the process of all of your multilayer checks and take a look at the documentation, something along the way isn’t going to make sense,” Lord said.

He advises lenders to check multiple sources, including contact information, business characteristics and registration filings. When in doubt, pick up the phone. A fraudster may have good-looking documents but may not be able to answer business-specific questions when put on the spot.

Leveraging tech to detect AI fraud

Lenders should use multiple vendors to detect fraud, with multilayer checks throughout the application process, Lord said. And those vendors should be using technology, as well as human intelligence and intuition, to spot scams.

“It really starts by being strategic about the choices you’re making around vendors,” Lord said. “Make sure you identify what it is that a human should be doing when they’re looking at [an application] and leverage technology to do some things where it can be easy for humans to maybe miss the mark.”

As AI capabilities grow, its use in fraud will grow, too. Lenders should remain familiar with the AI tools that can be misused and the tools that can combat fraud, he said.

“I certainly hope that all these people at OpenAI and Anthropic and Microsoft certainly do the right things and have safeguards in place, but you can’t necessarily rely on that,” Lord said. “Make sure that you’re prepared.”

Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.