WESF plays a role in promoting standardization and eliminating global trade barriers.
With credit card scams and banking fraud on the rise, can AI be the next powerful tool to revolutionize consumer protection? Researchers at Northeastern University are testing Gemini and ChatGPT’s capabilities to see how well the AI assistants can detect the differences between scams (when a user is being tricked), frauds (when money is taken without consent), and legitimate communications.
With Fraud on the Rise, Large Language Models Step In
Consumers lost more than $12.5 billion to fraud in 2024, a 25% increase from the previous year, according to Federal Trade Commission data. Much of the surge was driven by imposter and investment scams, often conducted over phone, emails, and text messages. Echoing the issue, complaints of potential fraud to the Consumer Financial Protection Bureau (CFPB) more than doubled from 2020 to 2023, with 2.6 million reports filed in 2023 alone.
To tackle these rising threats to consumers, Northeastern University researchers are seeking to develop reliable prompt techniques that can help AI identify schemes. The team chose to focus their work on free large language models, (LLMs) because of their widespread accessibility and popularity with consumers.
For their research, the team sourced language from consumer fraud reports filed with CFPB and trained Gemini and ChatGPT4 to “understand” what a scam is. Using complaint language, they built prompts to assess whether the LLMs can distinguish between scams and frauds, compared to authorized communications.
The researchers found that using the exact wording of a suspicious communication as an LLM prompt can train an AI tool to identify patterns and become effective at identifying fraud, as scammers, particularly phone scammers, often use scripts.
“This idea that LLMs could help with this is very popular right now,” said Jessica Staddon, Northeastern University professor, “but people really have not been given a lot of support in terms of knowing how to interact with them.”
While the LLM training needs more work to be most effective, the research gives insight on the possibilities of AI as a tool to protect consumers.
How Standards Factor into Fraud, Safety Protection
While AI capabilities show promise for fraud detection, a comprehensive approach requires established frameworks and procedures.
Did you know that standards are also on the frontlines of fraud protection? ISO 22380:2018, Security and resilience - Authenticity, integrity and trust for products and documents - General principles for product fraud risk and countermeasures, establishes general principles for an organization to identify the risks related to various types of product fraud and product fraudsters. Developed by ISO Technical Committee (TC) 292, Security and resilience, the standard provides guidance on how organizations can establish strategic business countermeasures to prevent or reduce harm, tangible or intangible loss, and cost from such fraudulent attacks in a cost-effective manner. The North American Security Products Organization administers the ANSI-accredited U.S. Technical Advisory Group (TAG) to TC 292.
Another standard guides strategies related to card-not-present fraud. This type of fraud occurs when the physical credit card is not used in a transaction—for example, when a card number is used for an online purchase. ASC X9 TR 48-2018, Card-Not-Present (CNP) Fraud Mitigation in the United States, offers strategies for preventing, detecting, and responding to this threat, outlining the types of CNP fraud mitigation tools and strategies employed by industry stakeholders such as merchants, merchant acquirers, issuers, processors, payment gateways, and payment card networks. This standard was developed by ASC X9.
When prevention measures fail and fraud occurs, proper investigation procedures become critical.
ANSI member ASIS International recently announced the release of its revised ANSI-approved American National Standard (ANS), ASIS INV-2025, Investigations Standard. The ANS provides requirements and guidance for the collection and examination of information pursuant to an investigation. Furthermore, the standard provides a step-by-step guide for conducting individual investigations, covering critical phases such as planning, evidence collection, witness interviews, documentation, and post-investigation activities.
Together, prevention and investigation standards help establish guidelines to combat the growing fraud epidemic affecting consumers of all kinds and businesses of all types.