AI, private banks' ultimate solution for KYCs?

I recently exchanged with a company offering innovative AI solutions so they can leverage my Private Banking (PB thereafter) understanding to determine how their technology further support PB processes. Following our chats, I informed them that I would use some generic parts of our discussion, especially on KYC (Know Your Customer) related topics to create a dedicated LinkedIn article.

Over the last few years, regulators took measures which translated in a significant increase of the weight, focus, relevance and criticality of KYC related topics (on-boarding, periodic reviews…). As a result, its share in financial companies’ overall workbook kept increasing, while topic related consequences of process failures or quality issues become heavier (fines, loss of licenses…) ?

As a reminder, KYC procedures are identification and documentation standards based on local and international regulatory requirements in order to increase monitoring and oversight on fraud, money laundering, terrorism support and other illegal financial activities. To that extent, financial institutions a required to have sufficient knowledge of their clients (identity, network, business activity, etc.) and keep monitoring it. Practically, it translated in client facing employees creating the historical summary of their prospects’ situations (family, source of wealth, businesses activities, etc.) during the onboarding phase and keeping it updated with material changes of situations, transactions and interactions over the life of the relationship under compliance’s supervision.

?

On the client side, multiple interactions for inquisitive questions, numerous forms filling and banks’ lead time to process completion the process often make it a painful experience. On top of client case complexity which understandably lengthen the time-to-yes (as it triggers internal discussions for conditions setting), the onboarding lengthy durations is mostly linked with bank’s manual processes. We hence spent most of our discussions on how artificial intelligence could support banks KYC processes the first 3 topics will tackle, while -though an AI enthusiast- I still kept in mind risks and limitations associated with the technology, that the 4th bullet will tackle:

1.?????? Automation & efficiency opportunities

2.?????? Data management and quality enhancements

3.?????? Client experiences improvements

4.?????? Challenges and question marks regarding privacy, acceptance, integrity and security.

?

AI offers significant upside for process automation and improvements:

One of AI’s potential contributions to KYC improvement lies with automating and enhancing its process steps, reducing the burden on human staff, ensuring regular updates and improving the overall compliance workflow. Below are the most important topics where in my view AI could be beneficially integrated into KYC processes:

a)Automation of onboardings and periodic reviews (incl. regular KYC updates as the document should reflect ongoing clients’ activities)

Onboardings documentation is driven by the bank’s staff (forms filling, documents submission… )but also require input from clients obtained through in-person meetings, which they often consider as intrusive.

A few topics I shared where I think AI could support onboardings and improve the customer experience are:

• Documentation through biometric recognition: AI can enable authentication through facial recognition, allowing clients to verify their identities remotely. For them, this potentially eliminates the need for a dedicated branch visit person making the process easier (it also ensures a certain consistency and takes away the risk that a negligent employee does not complete this formality thoroughly).
• Document authentication & verification: On top of identity scanning which we mentioned above, AI can verify identities against official documents (i.e. passports, national IDs, and driver’s licenses), by scanning and analyzing the images for authenticity. Optical character recognition (OCR) and deep learning algorithms allow systems to extract relevant data from the documents, compare it with submitted pictures to flag any discrepancies.
• KYC data gathering: AI can automate client risk profiling by analyzing data sources (newspapers, public records, social media, watchlists, etc.), detect red flags and highlight them for employee assessments (i.e. connections to politically exposed persons (PEPs), sanctions lists, or adverse media. This could drive a reduction of the workload and enable risk and compliance employees to focus on the part really adding value for the bank, i.e. the assessment of the red flags.

?

b/ Ongoing Monitoring and Reporting

AI could also play a significant role supporting that KYC are always kept up-to-date (KYC is not a one-time process and requires ongoing monitoring and updates). Leveraging the setup put in place for the initial data gathering, AI could also be setup to proactively monitor and highlight red flags to be reviewed. This would improve the efficiency of ongoing KYCs and reduce the burden of manual reviews.

• Ongoing monitoring of client behaviors: AI can track client activities in real time, monitoring for any changes in behavior that may indicate an increased risk, (e.g. unusual / unexpected large transfers or alterations in a client’s financial profile). AI can also flag accounts that have been dormant for an extended period, prompting further review to ensure they are still compliant with KYC regulations.
• Automated reporting: AI systems can automatically generate reports required by regulatory bodies, such as suspicious activity reports (SARs) and other compliance documents. This could reduce the time and efforts required by manual compilation while ensuring that banks remain compliant with regulatory deadlines.
• Regulatory watch: AI can monitor global regulatory changes and automatically update internal compliance systems to reflect new rules and requirements. This ensures that private banks stay ahead of evolving regulatory frameworks without needing to manually track and adapt to each change.

?

AI can play a major role in data quality improvements

Now focusing on key / complex clients, data quality is absolutely critical and AI could be supporting in several ways:

a/ Enhanced Customer Due Diligence (CDD)

Customer Due Diligence (CDD) is a critical part of KYC processes that involve enhanced and deeper scrutiny of high-risk clients or having complex financial profiles. AI can help automate those processes by leveraging big data analytics and machine learning models to gather and analyze relevant client information, enabling banks to assess risks more effectively.

• Advanced data analytics: AI systems can analyze vast amounts of data from multiple sources (e.g., financial transactions, credit reports, and media outlets) to detect unusual patterns or suspicious activities. This could be a great support for client facing employees, risk or compliance departments to identify potential signs of money laundering, fraud, or other illicit activities without manual intervention. AI highlighted cases, would then be assessed and commented by the employees.
• Risk scoring automation: AI algorithms can automatically assign a risk score to clients based on a combination of factors, such as their financial behavior, location, occupation, and historical data. This provides banks with a data-driven independent framework to assess clients risk levels and then dedicate the human resources accordingly.

b/ Anti-Money Laundering (AML) Compliance

AI is playing a crucial role in strengthening Anti-Money Laundering (AML) practices in private banking by automating and optimizing key AML compliance activities, including transaction monitoring, suspicious activity reporting, and ongoing client screening.

• Transaction monitoring and suspicious activities detection: AI can automatically detect patterns of suspicious activity by analyzing large volumes of data in real time. For example, machine learning algorithms can identify patterns of structuring or layering (methods used in money laundering) and flag them for further investigation. These algorithms are continuously refined based on new data, improving their ability to detect emerging trends in money laundering.
• Watchlist screening: AI systems can be integrated with external global databases, including sanctions lists, PEP lists, and law enforcement databases, to conduct ongoing screening of clients against known risks. AI can instantly cross-reference new clients or transactions against these watchlists and alert compliance teams if there is a match, reducing the chances of overlooking high-risk individuals.
• Predictive analytics for risk profiling: AI-powered predictive analytics can help banks better understand and predict potential money laundering risks by analyzing historical data, patterns, and behavior. This helps to assess the likelihood of a client engaging in illicit activities and ensures that additional due diligence measures are applied where necessary.

?

?

AI as a mean to improve client experience during the onboarding phase

AI-driven KYC processes can significantly enhance the overall client experience in private banking by making interactions faster, more convenient, and less intrusive. For instance:

• Seamless Onboarding: AI allows for faster client onboarding by automating document verification and reducing the need for face-to-face meetings. Clients can verify their identities and complete KYC requirements remotely and in real time, increasing client satisfaction and reducing friction during the onboarding process.
• Personalized Services: AI can analyze client data to provide a more personalized banking experience, including tailored financial advice, investment recommendations, and customized product offerings. By understanding client preferences and behaviors, AI helps private banks provide services that are better aligned with client needs and expectations.

?

?

Challenges in Implementing AI in KYC Processes

While AI has the potential to transform KYC in private banking, its implementation comes with several challenges and risks of its own:

1. Data privacy and security: AI relies on large volumes of data to function effectively, and this data can include sensitive personal information. Ensuring that this data is securely handled, stored, and processed is a critical concern. Private banks must adhere to stringent data privacy laws, such as the General Data Protection Regulation (GDPR) in Europe, to avoid legal and reputational risks.
2. Client trust and acceptance: Clients may have concerns about the use of AI in their KYC processes, particularly regarding the privacy and security of their data. Ensuring that clients understand how their data is being used and implementing transparent AI systems is essential for building trust.
3. Bias, neutrality and fairness: AI systems are intelligent but they can only go as far as the source data it is provided, in case of inaccurate or biased source’s data (i.e. yellow press, politically oriented sources…), AI’s output will similarly oriented output and /or contain the same flaws. This could lead to risk assessments inaccuracies or to unfair client treatment. To that extend, AI models should be assessed for fairness and transparency and their output pre-tested in a sample vs. manual reviews outcomes (to ensure consistency, accuracy, neutrality and faster / superior results)
4. Regulatory compliance: Regulatory requirements are constantly evolving and require human monitoring to ensure that AI frameworks are adjusted accordingly and that AI systems are updated to incorporate law changes, compliance standards updates and new regulatory requirements related to KYC, AML.
5. Cost of integration with existing systems and existing IT environment interaction: Many private banks have their own systems regarding KYC processes. Integrating AI-driven solutions in an existing (and potentially not that recent) IT infrastructure, can require significant cost investments.

?

Conclusion

AI could significantly be enhancing KYC processes in private banking by automating key tasks, simplifying and improving reviews, as well as streamlining the overall client onboarding. With AI-powered document verification, risk profiling, and transaction monitoring, private banks could increase efficiency, reduce human errors, and enhance their ability to detect suspicious activity taking care of a part of data gathering and allowing employees focus more on the assessment part. While there are challenges in adopting AI for KYC, including data security, regulatory compliance, and system integration, the benefits seem clear to me. I’d be interested to know ?there are different views let me know in the comments.

?