Like nearly everyone else, anti-fraud professionals are anxious to put artificial intelligence (AI) to work – especially generative AI (GenAI). In fact, 8 in 10 (83%) of fraud fighters expect to add GenAI to their arsenals within the next two years, according to the 2024 Anti-Fraud Technology Benchmarking Report by the Association of Certified Fraud Examiners (ACFE) and SAS, based on survey insights from nearly 1,200 ACFE members.

Enthusiasm for the technology is palpable, but questions remain: When will we see productivity and efficiency gains from GenAI? What are the best business cases for this technology? And how much will it cost to successfully deploy?

Such questions are giving many organizations pause. They have real concerns over regulations, data privacy, governance, and trust and ethics, to name a few. 

However, AI technologies deliver plenty of benefits. Benefits that organizations are realizing – right now. And while many business leaders stand on the AI sidelines, fraudsters are already capitalizing on GenAI and other advanced technologies. Unlike the lawful businesses, criminals don’t worry about regulations or bias, and they certainly don’t care about data privacy. That in itself is incentive for organizations to get started.

What – and what for – is GenAI?

A subfield of artificial intelligence, generative AI consumes existing data, learns from it, then generates similar output: text, images, audio, video or computer code. A disruptive technology, generative AI’s potential impact has been compared to that of electricity or the printing press.

Conversational AI models have rocketed in popularity among business and everyday users. The global market for GenAI is expected to soar to $1.3 trillion by 2032.

Long story short: Generative AI is here to stay. With that in mind, how can organizations begin to use GenAI to combat fraud? And conversely, how are criminals abusing it? Let’s first explore a few use cases for fraud fighters.

Spotting suspicious financial transactions

Financial firms have mere seconds to differentiate fraudulent transactions from legitimate ones. Thwarting fraudsters while delivering a friction-free experience to good customers is a tricky balancing act.

AI techniques, including adaptive machine learning and unsupervised intelligent agents, can enable banks to predict fraudulent transactions in real time – and reduce false positives – based on changes and inconsistencies in customer behavior patterns. Such real-time monitoring capabilities can help banks cut fraud losses, while reduced false positives can boost customer satisfaction, protect revenue, and lower costs.

Banks can also use GenAI during customer onboarding to scrutinize customer data and financial histories to assess credit exposure. This could help analysts make better lending decisions and curb loan defaults.

Helping tax agencies do more with less

Generative AI has the potential to create tremendous value for tax and revenue agencies. Applied to the right use cases, it could enable process automation, enhance decision-making, improve compliance and enhance taxpayer experiences. 

For example, using GenAI to augment rote functions like reviewing and responding to taxpayer documents could dramatically boost efficiency. Many such documents, like identity theft affidavits submitted by taxpayers, include open-ended text fields that require intensive manual review. The result: backlogged cases waiting to be examined, classified and followed up on.

GenAI models can automatically extract meaning from documents, summarize information, and help examiners determine next steps. Its application could help expedite review queues and reduce staff time spent per case.

Protecting the integrity of the procurement process

Sourcing and procurement operations have historically been at the forefront of technological disruption. From using advanced analytics for spending categorization to deploying conversational AI for guided buying, source-to-pay tools have continuously innovated to address process challenges. Yet many sourcing and procurement functions struggle to optimize efficiency and manage risk and costs.

A recent survey of chief procurement officers (CPOs) by Deloitte found that more than 70% believe procurement-related risk and supply chain disruption has risen over the past year. Risk evaluation tools need capabilities to continuously monitor external risk factors, ingest voluminous data, and perform advanced analytics to predict and prescribe risk key performance indicators and preventive management. Though cost management has always been the CPO’s focus, rising inflation has put additional pressure on procurement organizations to further optimize costs.

Generative AI can help address procurement challenges by:

• Crunching large sets of data to process scenario-based results, reducing complex manual processes and interventions.
• Using complex automation to increase efficiencies.
• Generating actionable insights based on historic trends, demand profiles and supplier performance.
• Combining internal and external data to craft better negotiation strategies.

GenAI can be used in procurement compliance management to monitor procurement processes and identify potentially fraudulent activities and anomalies. Additionally, trained on insights from historical noncompliance, AI systems can learn to recognize similar patterns in the future.

GenAI: Fighting fire with fire

You may not see them, but fraudsters are always lurking, waiting for just the right moment or weakness to strike. They have the skillset and are primed to forge any innovation into their shiny new sword.

GenAI is no exception. Tools like FraudGPT and WormGPT are arming criminals with novel ways circumvent security. Who is vulnerable? Everyone. SAS’ recent Faces of Fraud study, based on a survey of 13,500 consumers in 16 countries, revealed that:

• Seven in 10 have fallen victim to fraud at least once; 40% have experienced fraud twice or more.
• A whopping 89% believe organizations should be doing more to protect consumers from fraud.
• Two-thirds would switch providers due to a fraud experience or if another provider offers better safeguards.

These findings emphasize the importance of organizations’ anti-fraud preparedness. GenAI tools are helping bad actors create more persuasive phishing emails, more convincing phone-based impersonation scams, and more imperceptible malware. In this new reality, fraud fighters need defenses of the same caliber. They must fight fire with fire, GenAI vs. GenAI.

Getting started

Ready to take the generative AI plunge? No need to leap from the high dive. Consider a measured, deliberate wade into the shallow end of the pool to start.

• Choose the right partners. Despite fervent interest by fraud examiners, the ACFE and SAS found that AI and machine learning adoption in anti-fraud programs has grown only 5% since 2019, climbing from 13% to 18%. This AI adoption gap – explored in this Banking Exchange Q&A with SAS’ Stu Bradley – highlights the complexities of AI deployment. It also underscores the importance of choosing the right technology and professional services partners to streamline adoption and build a foundation for success.
  
  AI and GenAI aren’t simple plug-and-play technologies. Often times their dividends can be more readily attained by deploying modularized solutions on a single, cloud-based platform. This way the organization can reap rewards in one area (fraud detection, for instance) and gradually expand to other facets of the business (risk management and marketing automation among them), using shared data on a single platform.
  
  
• Build momentum with an early win. Many organizations make the mistake of adopting AI for AI’s sake, neglecting to align with business priorities. AI that doesn’t deliver production value is AI wasted. Start with a focused pilot program to show positive results, like automating a manual process for greater efficiency. According to the ACFE and SAS’ recent study, 68% of surveyed anti-fraud pros indicated that lack of perceived ROI was a moderate to major challenge in adopting new technologies.
  
  An out-of-the-gate success that delivers on AI’s promised benefits can drive support and budget allocation for future business cases with senior executives. For example, an insurer who first deploys GenAI to automate claims processing could later use it to detect potential fraud, support underwriting and/or enhance customer relationships.
  
  
• Always keep the human in the loop. As organizations use AI technologies to automate an ever growing number of decision-making functions, the criticality of responsible innovation cannot be overstated. Business and government leaders have an obligation to wield these advanced technologies in ethical, human-focused ways. That requires establishing the appropriate guardrails. As a baseline, every GenAI use case should involve meaningful, substantive human review of the data inputs and any generated content and outputs, scrutinizing them for accuracy, potential bias and quality issues.

Make no mistake, generative AI will change the world. It already is – for better and for worse. In the right hands, GenAI can deliver progress, productivity and efficiencies. In criminal hands, it has the potential to wreak havoc to the tune of untold billions in fraud losses.

Who will win the GenAI arms race? The outcome likely depends on the choices we make today.