Financial institutions must use AI to counter the rising number of AI-driven scams that resulted in a global financial loss of $1.03 trillion in 2024.

Authorized push payment fraud is the most prevalent type of fraud in the United States and the United Kingdom, according to a report published today by venture capital fund Team8 referencing data from the Global Anti-Scam Alliance.

Credit unions suffered major losses with 33% reporting a 50% to 100% spike in scam incidents in 2024, Ronen Assia, managing partner at Team8, told Bank Automation News.

The Consumer Financial Protection Bureau in December 2024 sued JPMorgan Chase, Bank of America and Wells Fargo for failing to protect consumers from fraud via the Zelle peer-to-peer payments app in 2017, resulting in consumer losses of more than $870 million.

Gen AI will only lead to more prevalent, costlier fraud, said Assia.

“With the rapid advancement of generative AI, deepfakes and real-time payments, this trend is likely to continue, driving both fraud cases and financial losses even higher in the coming years,” Assia said.

“Generative AI is rapidly scaling finance and identity fraud, enabling highly personalized attacks with just a click,” he added.

Imminent threat

Gen AI-powered tools such as FraudGPT, which uses advanced AI models to generate synthetic content are enabling scammers to commit automated payment fraud, Assia said.

Phishing attacks, identity theft and AI-generated phishing emails are also becoming nearly indistinguishable from legitimate messages, Assia said.

Social engineering, which refers to a manipulation technique that tricks people into performing compromising actions or divulging confidential information that aids fraudsters, is becoming more advanced through gen AI-powered video and audio deepfakes, posing a challenge for financial institutions trying to verify a customer’s identity, he said.

Tom Tovar, chief executive at software company Appdome, agreed, saying AI can be used to create fake identities for fraudulent loans, credit applications and money laundering.

“Deepfake fraud has become a significant concern for businesses, especially in the financial sector,” he told BAN.

The evolution of AI-powered security threats began last year, he said. “In the past four months, AI deepfakes created an existential crisis in facial recognition,” he said.

“We are at the beginning of the war on AI-powered fraud. It’s evolving minute by minute.”

— Tom Tovar, CEO, Appdome

Gen AI for fraud protection

Fraud prevention is the third most common use case among banks for AI deployment and the second among credit unions, Team8’s Assia told BAN, referencing a January report by Cornerstone Research, which surveyed more than 300 banks and credit unions.

Banks must apply AI to all facets of a defense lifecycle to remain proactive against escalating AI-driven fraud, Tovar said.

“The fundamental reality is that the only way to fight AI is with AI,” he said. “If banks, financial institutions and fintech platforms don’t embrace AI-driven protection, the power of AI-generated fraud will outpace them by the end of this year.”

Banks with assets of $50 billion or greater are investing the most in internal gen AI deployment, including anti-fraud measures, Assia said, adding that small and medium-sized FIs that lack internal resources can outsource AI capabilities to specialized fintech vendors.

JPMorgan, which has used gen AI-powered large language models for payment validation screening since 2021, has boosted the speed of account validation by 15% to 20% through gen AI deployment, according to the $4 trillion bank.

Meanwhile, since adding AI to its decisioning tool a year ago, Mastercard has increased its fraud detection rate by as much as 300% in certain scenarios while reducing false positives by more than 85%, Ken Moore, chief innovation officer at Mastercard, recently told BAN.

Gen AI provides real-time analysis and pattern recognition, enhancing fraud detection capabilities, according to the Team8 report. AI algorithms also process and analyze data much more quickly than traditional methods, which is critical for FIs that handle millions of transactions daily, according to the report.

Appdome ups its AI defense

Appdome added 30 new defense plugins to its gen-AI powered Account Takeover Protection suite for deepfake detection in mobile apps, the company announced today.

The plugins protect app data by detecting:

• Deepfake videos;
• Voice cloning; and
• Facial recognition spoofing.

Appdome’s suite is already seeing widespread use by financial institutions in 2025, Tovar said, without disclosing client names.

As banking becomes increasingly digitalized, financial institutions need to tackle fraud detection where it starts, not just where it ends, Tovar said.

”That means bringing anti-fraud, know your customer, enhanced due diligence, anti-scam, social engineering, geo-compliance, deepfake detection, etc. into the mobile app itself.”

Register here for Bank Automation Summit 2025, taking place March 3-4 in Nashville, Tenn. View the full event agenda here.