Fraud and Artificial Intelligence: Potential developments and threats to watch

Artificial Intelligence (AI) and Language Models (LLM) have been dominating the news lately with one tool in particular: ChatGPT. AI can be a tremendous tool of public good and knowledge, particularly in combating digital fraud and other types of financial crime. In fact, more and more organizations are using it to digitize their know-your-customer (KYC) processes and ensure that the person on the other end of the transaction is who they say they are.

These tools can automate the analysis and processing of user data, personal information, transaction history, and remove internal bottlenecks, enabling scalability at a smooth pace.Prev. They can also check users for sanctions and politically exposed persons (PEP) lists and report potential fraud and suspicious activity in real time.

However, history has taught us that with every innovation and technological development, it does not take long for a new criminal initiative to appear to be exploiting potential loopholes in new regulations or legal loopholes. Recent advances in AI and LLM are unfortunately no exception. Here are five things to know about AI fraud.

1. Facilitate and accelerate social engineering

The increase in productivity offered by ChatGPT and other similar tools has generated widespread enthusiasm, but their use for fraudulent activities is unfortunately inevitable. Scammers using impersonation and social engineering have already begun to use AI for “romance” and “phishing” scams by producing fake documents such as invoices, contracts, and tax documents, which are often more personalized and accurate than those from the sources they are impersonating. It is important to watch and look out for details: fake documents may contain spelling or grammatical errors that may make it easier to detect their fraudulent status. But thanks to the power of artificial intelligence, errors in a fraudulent document can now be corrected, making it difficult to distinguish between a genuine document and one produced with dishonest intentions.

2. Cybercrime is less technical

It used to be that scammers and seasoned cybercriminals had a good knowledge of software and code to produce documents or websites that would trick innocent people into handing over their hard-earned money. Today, criminals with basic computer skills and no programming knowledge can engage in cybercrime thanks to the simplicity of AI tools. They can quickly and easily create malware, VBA (Visual Basic for Applications) programs, and virtual chatbots that attract victims.

3. Simplify attempts to falsify identity documents

LLM is not the only rapid development in the field of artificial intelligence. The tools to create compelling images, videos, sounds, and 3D models are increasingly available to the general public. This means that fraudsters will soon be able to create ultra-realistic documents, such as passports and other official documents, that will even contain the traditional seals currently used to authenticate real identity documents. Identity theft and obfuscation strategy will become much easier because these generative AI models will allow the fraudster to perform realistic attacks against identity documents and the biometric information they contain. there

Document verification, and in the case of deepfake technology, life detection is likely to create problems for companies that don’t have these kinds of controls and the more rigorous facial recognition technologies in place.

4. The privacy of chat data is threatened by existing loopholes

One of the many shifts in ChatGPT’s history occurred in March 2023, when the Italian data protection authority shut down the platform in their country. The reason given was that it contained a bug that made users’ chat history and personally identifiable information (PII) visible to others, allowing it to be used for fraudulent purposes. Personal data such as names, passwords, addresses, and anything else a user can put into LLM can be used to impersonate people online and perpetrate artificial fraud.

5. Access to reliable and truthful information will be more complex

Although not directly related to fraud issues, the veracity of information will be increasingly prioritized as the development of AI and master tools progresses. The risk of “fake news” is likely to increase exponentially with the amount of content that appears original. This warning should also apply to database checks. Businesses that need to perform KYC and Anti-Money Laundering (AML) checks are advised not to rely solely on an LLM to do them. Responses to sanctions and political politicos can be easily manipulated leading to inaccurate results.