A recurring theme when I talk with banking leaders about the management of compliance records, contracts and other critical documents is that it is both essential and expensive work. Bank attorneys, paralegals and loan officers spend thousands of hours poring over countless pages of regulatory compliance filings, loan agreements and other records to determine whether they comply with laws, terms and conditions. Banks spend an estimated $70 billion annually on regulatory compliance and governance software, a figure expected to grow to nearly $120 billion by 2020.
What can be done to rein in the cost of regulatory compliance? By unleashing artificial intelligence (AI), banks may be able to save an estimated 30% of compliance costs while accelerating throughput up to three times.
The use of AI – particularly natural language understanding (NLU), a subset of natural language processing (NLP) – can have a major impact on compliance efficiency and robustness. Already, automated review systems built on machine learning technology are generating stunning reductions in the time and effort required for document interpretation, with fewer errors in the process.
We have seen this in work we did with a major mortgage solution provider to automate the recognition and extraction of critical data from multiple sources, reducing costs and the manual effort required for compliance updates.
Applying AI to Compliance
AI is helping regulatory compliance teams interpret regulatory meaning, comprehend what work needs to be done and codify compliance rules. The COIN (Contract Intelligence) program created by JPMorgan Chase, for example, reviews documents based on business rules, data and data validation, flagging potential issues. In seconds, the system can examine loan documents that would ordinarily consume 360,000 hours of work.
In my experience, areas in which AI can contribute to compliance efficiency and effectiveness include:
- Know your customer. AI’s ability to analyze vast amounts of data, scrape the web and find patterns can strengthen and streamline KYC processes. Algorithmic machine learning models with a layer of human-like cognitive reasoning can raise red flags, an increasingly important capability with the broad spectrum of technologies in play.
- Money laundering detection. Monitoring reports, news items and regulatory alerts can be evaluated as risk indicators using AI, and those indicating the greatest exposures can be subject to further analysis.
- Rogue employee detection. AI software could catch employees opening fake accounts by finding multiple accounts using the same e-mail or IP address.
- Trader monitoring. Using AI to learn trader personalities and behavior can increase the precision of suspicious trading, helping avert costly false alarms.
Streamlining Regulatory Response
In addition to strengthening safeguards, AI can help banks apply and stay abreast of regulatory requirements:
- Law and regulation parsing. NLU software can select specific rules in lengthy regulatory documents and send them to people and departments that need to comply.
- Identification of regulation and policy updates. Cognitive engines now available can understand and analyze high volumes of regulatory changes and verify that a business is alerted to the most up-to-date policies.
- Identification of units, products and processes affected by compliance requirement NLP systems can analyze documents to identify people, products and processes affected by legal and regulatory changes.
Along with these uses, AI’s contributions to regulatory compliance will continue to evolve and expand in the future. For example, data analysis and visualization tools will continue to create richer views of big data, provide the ability to generate meaningful reports for regulators and be used internally to improve key business decisions. Also, advanced algorithmic machine learning models will raise red flags with a layer of human-like cognitive reasoning, strengthening KYC capabilities and compliance.
Overcoming AI Concerns
These opportunities to improve regulatory compliance with AI come with some caveats. For one, AI contributes to the growing risk of fake and unverifiable news affecting algorithmic trading. Also, automating decisions based on known conditions can detach compliance professionals from decision-making processes and create false confidence in the comprehensiveness of regulatory monitoring.
Growing tension over AI source code is another concern. Some regulatory authorities suggest that codes be handed over to regulators to fight crime, while financial institutions worry whether their intellectual property will remain secure.
Despite these challenges, AI offers huge potential to improve compliance and help banks stay relevant with customers and partners. And as their competitors continue to adopt AI solutions, banks will have no choice but to follow suit.
By replacing legacy systems with advanced systems of intelligence powered by AI, financial institutions can substantially reduce the cost of compliance, increase reporting accuracy and reduce the risk of financial damage related to inappropriate risk assessment.