Feedback from technology providers and vendors on the use of artificial intelligence (AI) and other automation tools in response to a request for information issued by the Office of the Comptroller of the Currency (OCC) last year, offers a glimpse into the responses regulators might expect from a similar request about the use of advanced technologies issued last week.

Last week’s request for information was issued jointly by the OCC, the Federal Reserve, the Federal Deposit Insurance Corporation (FDIC), the Consumer Financial Protection Bureau and the National Credit Union Administration (NCUA). The request includes a total of 17 questions and focuses primarily on technology, whereas the 2020 request focused only partly on artificial intelligence and machine learning.

Additional guidance requested

Most of the 2020 submissions from technology providers requested additional guidance from the OCC on the use of advanced technologies, with providers arguing that many banks are shy to adopt AI, machine learning (ML), and other tools because of potential regulatory fallout. Several vendors cited a need for OCC guidance on third-party tech providers to create what Plaid called “a level playing field for small banks’ data access.”

That said, vendors also called for a principle-based approach rather than a potentially more restrictive, specific approach to rules, with some urging the OCC to create a sandbox for exploring different technologies.

Google

Google provided several pages of comments addressing everything from trends driving financial institutions to adopt cloud computing to a list of recommendations for designing real-time payment systems, with recommendations for hack-a-thons and other innovative practices.

Google didn’t mince words, however, regarding AI and the automation of anti-money laundering (AML), combating the fighting of terrorism (CFT) and know your customer (KYC) measures.

“The technology holds particular promise in improving AML/CFT compliance and reporting, KYC processes and fraud detection and prevention,” Google commented. “With respect to AML/CFT, current heavily manual and labor-intensive approaches to identifying and combating financial crimes have not been effective.”

Despite banks spending billions, 95 percent of security alerts are deemed false positives, Google noted. Banks have also faced hefty fines for failing to properly identify activity that should have been flagged, including drug trafficking, human trafficking and terrorist financing. This, according to the tech giant, is a technology problem.

Rules-based technology is ineffective in identifying true crime while generating substantial false positives; by comparison, AI can develop “more sophisticated analytical lens” in a timely way and detect new patterns that bypass rules-based logic, according to Google. While it applauded the OCC and other federal agencies for taking some steps to support SAR automation, it noted that the regulator could go further by outlining the parameters for using machine learning tools.

“For both KYC and fraud-detection applications, increased regulatory clarity regarding expectations and parameters for when a bank is permitted to rely on machine learning tools and related automation is critical,” Google noted.

Plaid

Plaid provides integration for data sharing between banks and apps, so it’s not surprising that its response focuses on data availability for AI, as well as data sharing between smaller banks and third-party providers, using APIs to consume and share data.

With regard to AI, Plaid focused its comments on the use case for consumer lending, stating that AI can provide more complex analysis of data related to rent, utilities and other bills to better predict credit risk. The San Francisco-based company pointed to studies showing how AI and ML use could effectively cut bank credit losses and increase the percentage of accepted loan applications.

Plaid urged the OCC to support data sharing by “encouraging banks to enhance their internal data capabilities as a means of improving efficiency, and by insisting on a level playing field for all banks with regards to data sharing.”

The company also asked the OCC to clarify a March 2020 updated FAQ on third-party relationships, which Plaid said “may have inadvertently” sent a message that an API agreement with a third party might be less viable for smaller banks.

“API strategies are fundamental to banks’ digital futures, and the OCC should clarify that third-party vendor-based API relationships are not higher risk than legacy relationships,” Plaid wrote. “Otherwise, small financial institutions will hesitate to adopt formal API structures and could fall behind the advancements possible in digital financial services.”

Upstart

Upstart operates a platform that enables lenders to offer loans to consumers. It uses both AI and ML algorithms to more accurately price consumer credit, according to its statement. The fintech blamed a lack of regulatory clarity on the slow adoption of AI in credit underwriting.

“Despite these benefits, banks’ uncertainty about the specific model risk management and governance requirements that apply to AI models, particularly AI models used in credit underwriting, have impeded many of them from using the technology,” Upstart stated.

The problem, according to the provider, could be quickly addressed by guidance from an examiner that specifically addresses: the documentation banks need to maintain on models; the level of understanding they must have of how the model works; and how appropriate testing and validation are compromised at the bank level.

Upstart’s studies and external analysis have shown its credit underwriting model, based on AI and ML, is more accurate than traditional models and “allows for greater access to credit at lower rates of loss,” according to its comments. It also argued that the technology “can eliminate human unconscious bias in the credit underwriting process,” a remark that the Consumer Report response contested, pointing specifically to allegations of bias in Upstart’s model.

Organizations have until July 7 to respond to the OCC’s current request for information.

The regulator did not respond this week to questions from Bank Automation News about whether any changes have been implemented or are planned as a result of its July 2020 Advance Notice of Proposed Rulemaking.

Bank Automation Ignite, on April 13-14, is the event for inspiring automation initiatives and investment in financial services. At the virtual event, financial services professionals can discover new use cases and technologies that are accelerating automation in banking. Learn more and register at www.BankAutomationIgnite.com.