GenAI: friend or foe in fraud risk management?

Generative AI has taken the world by storm since the launch of ChatGPT in November 2022.

While advancement in tech is made with the best intentions to improve our lives, criminals can also use it to their own benefit

The technology’s ability to generate human-like conversations and impressive original content is being extolled by experts and common users alike across industries. From students to researchers, consultants, software developers, artists, media content creators, businesses, consumers of various services – the overwhelming adoption of GenAI by all in such a short time is testimony to its benefits and future potential.

But there is a risk that GenAI may be used by fraudsters to create extremely convincing messages, advertisements and other forms of content to steal money and identity information from unsuspecting consumers and enterprises. Can GenAI-powered innovation also aid in defending against such crimes and help businesses identify fraudulent signals, generate threat risks and develop stronger controls?

A tech revolution with boundless benefits

While still in its nascent stage, GenAI has shown a lot of promise with a broad range of utility across industries, from art and literature to media, music, retail, healthcare, banking and finance, travel and hospitality, manufacturing, e-commerce and several others. Its ease of use and ability to create human-like content – text, images, audio, video and even logical conversations – with simple prompts from users makes this technology versatile and efficacious.

Be it generating a brief of loan/credit proposals or summarising actionables from regulatory and audit documents for financial institutions, generating high-resolution medical images in radiology and pathology for enhanced diagnosis, creating personalised and optimum end-to-end travel itineraries, product design and supply chain optimisation in manufacturing – there are numerous high-impact uses of GenAI that can dramatically boost operational efficiency and customer engagement in almost every industry.

A friend of fraudsters, too

Technology has always been a double-edged sword. While advancement in tech is made with the best intentions to improve our lives, criminals can also use it to their own benefit – for committing frauds and other crimes. So with the overwhelming adoption of GenAI already evident among businesses and consumers, and with its usage only expected to grow, we can almost foresee how fraudulent activity will also get a boost using the same technology.

Deepfakes and voice cloning have already become fraudsters’ tools for scamming people by making them believe a call for payment is from someone they know. Social engineering scams are likely to rise as scammers can now use GenAI to create texts, advertisements, calls and emails more convincing, realistic and personalised.

In today’s world, the more alert and savvy customers can identify scam messages and mails through errors in the text composition (spelling and grammar), imperfect content (typos and sentence construction), the email ID format and so on. With GenAI, fraudsters are likely to create more professional, organised and realistic content that may be hard to find flaws in.

GenAI-powered text conversations through social media or ads in online marketplaces for example can lead to frauds by convincing users to make payments or share personal identity and payment details. Images and videos generated through GenAI closely resembling target victims can be used to open fraudulent accounts, as it might get increasingly difficult to distinguish between the real and fake during ID verification and KYC checks. Even novice criminals can now plan sophisticated scams by generating perfect codes and programmes for fraudulent attacks by prompting GenAI to write and edit them.

Protecting consumers from conmen: GenAI’s superpowers

For financial institutions (FIs), fighting increasingly complex and sophisticated frauds has assumed immense importance to protect consumers, communities and businesses. FIs have been exploring various new technologies to detect and prevent frauds and financial losses in real time.

And now we have GenAI, whose superpowers can potentially help FIs in this endeavour. Discovery of new fraud typologies on a continuous basis remains a challenge for FIs. GenAI can synthesise large volumes of unstructured data from external sources like web and news media, investigative journalism reports, regulatory exemption applications, shared fraud databases and so on to summarise new fraud patterns/risks and classify them into new fraud typologies.  Controls can then be designed accordingly to defend against these new frauds.

Building high-performance fraud detection algorithms today is dependent on real-life customer and transaction data to train and validate the models, which has remained a constraint. GenAI can help with realistic synthetic data creation for model training and validation, scenario and fraud attack simulation to identify vulnerabilities and design controls to mitigate these risks.

Customer due diligence (CDD) is a critical function in fraud prevention – be it new client onboarding or new credit approvals (loans, credit cards, increasing credit limits) for existing clients. GenAI can be a great tool to go through piles of KYC documentation and reference them with customer-filled forms and other subscribed data sources of the FI to come up with a CDD summary report.

GenAI can also be used to analyse user communications with FIs – such as emails, chats, documents and product and service requests – to extract insights on financial behaviour, sentiment analysis for intentions and potential risks of fraud. Fraud investigations can also leverage GenAI for alert and dispute resolution by accessing different sources of information on the context and providing a summary of the case that will aid in its decisioning.

GenAI for the greater good: a vision

GenAI is being touted as one of the greatest technology innovations since the launch of the internet 30 years ago. Can this technology beat the criminals at their own game by generating threat risk reports of a new loan or account origination application to help detect synthetic ID fraud? Can GenAI scour online marketplaces and social media and generate instances of FI impersonation and investment scams, which can then be taken down? Can GenAI differentiate between real and fake images, documents and biometrics (fingerprint scans and selfie videos, for example) during identity verification and payment authentication to prevent account takeover frauds?

Maybe not just yet as the technology is still evolving, but this reality may not be too far in the future as the industry is already exploring how GenAI can be used for the greater good, including customer protection and fraud prevention.

Exciting times are ahead, and I look forward to witnessing this vision unfold into reality.

About the author

Sujata Dasgupta is global head – financial crime compliance advisory at Tata Consultancy Services, based in Stockholm, Sweden.

She has over 20 years of experience, having worked extensively in the areas of KYC, sanctions, AML, and fraud across banking, IT services, and consulting.

She has worked with premier banks in several major financial hubs in seven countries across the US, UK, EU, and Asia.

She can be contacted on LinkedIn.