Artificial-intelligence image tools are powering a surge in fake receipts, tripping up corporate expense systems and finance teams. Expense-audit provider AppZen said AI-generated receipts made up about 14% of all fraudulent submissions it saw last month, up from none a year earlier. FinTech firm Ramp said software it deployed blocked more than $1 million in phony invoices in 90 days. About 30% of U.K. and U.S. finance professionals surveyed by Medius reported an uptick in fabricated receipts following OpenAI’s release of GPT-4o in 2024. “Do not trust your eyes,” said Chris Juneau, a senior vice president at SAP Concur, which uses AI to run more than 80 million compliance checks a month. Vendors say the latest image models can produce realistic wrinkles, itemization and signatures that closely mimic legitimate documents. OpenAI said it takes action when policies are violated and embeds metadata in images generated by its tools. The escalation is part of a broader shift in fraud, as criminals combine generative AI, voice cloning and deepfakes to impersonate executives and authorize payments. PYMNTS Intelligence estimates 68% of organizations encountered at least one attempt against accounts payable last year. Companies are leaning on automated controls and provenance signals, but the speed and accessibility of AI has lowered the bar for would-be scammers, raising the stakes for finance and risk teams.
Related articles:
Facing Reality? Law Enforcement and the Challenge of Deepfakes
NIST AI Risk Management Framework 1.0
