Financial fraud is an epidemic. The U.N. Office on Drugs and Crime estimates that global financial networks see up to $2 trillion in money laundering activity annually. Of that, only about 1 percent is intercepted. The low rate of detection leaves criminals with high odds of success in an activity that is often used to fund terrorist groups or aid criminal enterprises.
Money laundering is just one type of financial fraud. Hacked financial accounts, identity theft used to empty accounts or make illicit purchases, embezzlement, and tax evasion round out the full picture of the manifold threats to financial institutions. The U.S. Department of Treasury’s Financial Crimes Enforcement Network says $1 billion is lost to cybercrime monthly in the U.S. alone.
Traditional financial institutions are ill-prepared to deal with the increasing sophistication of financial cybercriminals, especially as the volume of transactions grows each year. That’s why many experts in this field are looking to two technologies to help stem the tide of financial fraud: AI and blockchains.
How can AI stop financial fraud?
Many financial professionals believe artificial intelligence (AI) will be a critical tool for better detecting crimes in progress and for preventing financial fraud before it happens. AI models can analyze patterns of behavior, with the ability to detect nefarious activity far faster and in more sophisticated ways than human professionals.
Through machine learning (ML) or deep learning, AI models can process huge quantities of data, acquiring more insights over time and become increasingly more effective at detecting fraud. A deep learning model trained to detect financial fraud can learn continuously from both the database of past purchases and from current activity to identify emerging patterns or unusual transactions that may signify fraud.
Activities an AI model can flag for investigation could include unusually large or small transaction sizes, many transactions in a short time from a single device, or multiple purchases from a single customer originating from different locations over a short time. AI can analyze customer behavior at the individual level, noting when a particular customer suddenly begins making large or unusual types of purchases.
Once trained, the AI model can operate independently to analyze patterns and identify possible problems, potentially offering both increased efficiency and tremendous cost-savings to financial institutions.
Using natural language processing (NLP), an AI model can analyze the content of emails, chat transcripts, or even social-media posts, making connections in ways that would be difficult for humans to quickly accomplish. For instance, if a customer alters their account details and then requests a password reset, an AI model could be trained to flag this as potential attempted fraud. The models can also rapidly adapt to emerging gambits by attackers.
Employing AI for fraud detection enables institutions to spot more complex fraud schemes that traditional detection systems may miss. For instance, an AI model would be more likely to quickly uncover a pattern of fraud that’s spread across different channels, such as a combination of in-person and online transactions made by one user to the same retailer.
The predictive analytics of a sophisticated AI model can also prevent fraud by identifying higher-risk customers or transaction types, alerting fraud professionals about types of fraudulent activity they should monitor more closely.
Combining AI and blockchain for fraud prevention
AI models and blockchain structures each bring advantages when it comes to detecting and preventing fraud. The benefits can be multiplied by bringing these tools together.
If you combine the cryptographic security of blockchain with the advanced analytical capabilities of an AI model that’s been designed specifically to secure a database, you have a more robust system to both repel attackers and detect financial fraud in real time.
Financial crimes don’t just affect banks. AI-based fraud prevention tools could help retailers detect payment fraud or insurance companies uncover healthcare fraud. Fraud professionals are often doing essentially forensic work, analyzing financial crimes after they happen for future prevention insights. AI models have the bandwidth to analyze large datasets in real time to identify suspicious behaviors as they’re happening, giving institutions the power to shut down crimes before more losses occur.
The challenges of AI and blockchain adoption in finance
If blockchain and AI offer such advances for preventing financial crimes, why aren’t traditional financial institutions all using them? The answer is complex. As we discussed in a previous article on AI for compliance, banks tend to be skeptical of tech innovations and to prioritize adding people over updating systems.
A recent study by Refinitiv of over 3,100 financial-industry compliance managers with an average of $33 billion in annual transaction volume dug deeper into the challenges of implementing AI and blockchain in the traditional financial-service industry. The study found that the need to adopt better technology is pressing: 72% of organizations reported their organization was aware of financial crimes in their system in the past year. At the same time, compliance with basic anti-crime steps was lax.
Only 53% said they consistently perform know your customer (KYC) checks on clients’ identity. And that was the best result the survey found for compliance measures: other basic steps such as KYC on customer data, screening for sanctions, screening of suppliers, and transaction monitoring all saw less compliance. Only 28% reported they screened for financial crime risk.
Nearly all–97%–of respondents believe advanced technology, including AI and blockchain, could help them stop fraud. For instance, AI offers the promise of far faster, easier KYC checks that might help improve verification rates and screen out bad actors at the start.
Still, obstacles to AI and blockchain use at traditional institutions remain formidable.
Less than half of these financial organizations–44%–currently use AI or blockchain. Over 80 percent said financial regulations are an obstacle to collaborating with other organizations to fight financial fraud. There’s also a lack of tech expertise at these organizations: 73 percent said they struggle to roll out new technology. Perhaps as a result, less than 40 percent said they’re able to monitor transactions in real time.
When it comes to AI and blockchain, many compliance managers aren’t yet believers. Only 56% thought AI would significantly help prevent financial crime. Even fewer, just 40%, saw blockchain as an important tool.
What will motivate financial institutions to implement AI and blockchain to help prevent and detect financial fraud? New regulations may be the top motivator, with 59% citing it as such. The need to prevent financial crime ranked slightly lower at 56%. Less than half of respondents believed adoption would cut costs, even though cost-savings has been one of the strongest cases for tech adoption.
Both AI and blockchain show great promise for preventing financial crimes–but they also pose challenges. AI isn’t foolproof–in fact, it’s only as good as the data it was trained on. So AI can produce false positives or negatives in scanning for possible fraud.
The future of AI and blockchain for financial fraud prevention
There’s tremendous potential for AI and blockchain technology to enhance fraud prevention and prevent financial crimes, especially when their advantages are combined. The promise of real-time fraud detection with AI could be a game-changer in stemming financial fraud losses.
It remains to be seen if traditional financial institutions will move to implement these technologies–or if the advantage will go to DeFi projects that seize on the capabilities of AI and blockchain to build more trusted financial environments.