Discover's AI use, Synapse's bankruptcy: Top tech news for April 2024

In this month's roundup of top tech news: The impact of Synapse's bankruptcy on the banking-as-a-service industry, large language models best suited for banks, Discover deploys generative artificial intelligence in its contact center and more.

Click here to read last month's roundup of top tech news.

Citigroup's Wave Of Job Cuts Poised To Start As Soon As Monday
In an ongoing case, Citi and the New York attorney general are disputing who is responsible for wire transfer fraud that drained a customer's retirement savings.
Nathan Howard/Bloomberg

Europe may have an answer to U.S. wire transfer fraud questions

Article by Carter Pape
A move by Citigroup to dismiss what it called a "misguided" and "imaginative" wire-fraud lawsuit by the New York attorney general has gotten a mixed reception among bankers, many of whom sympathize with Citi's pushback while others say banks can do more to protect their customers.

The move also highlights ongoing debates in the U.S. and abroad about who should be liable when a consumer loses money to a bank spoofing scam. While Europe is moving toward holding banks liable, the U.S. has not seen any such proposals.

Letitia James, the state's attorney general, sued Citibank in January for inadequate responses to "obvious red flags of identity theft and account takeover" cases, allowing fraud to take place, including against one customer who had $40,000 stolen from her via wire transfer after she clicked on a fraudulent link she received in a text message. Citi denied her case, according to the lawsuit.

Click here to read the full story.
Key Speakers At The Messari Mainnet Summit
The SEC has proposed a $2 billion fine on Ripple for its sales of XRP, which the SEC calls unregistered securities.
Michael Nagle/Bloomberg

Ripple's proposed $2 billion fine would close ugly chapter

BankThink by Penny Crosman
A final ruling expected in the Securities and Exchange Commission's lawsuit against Ripple this spring could provide some clarity to the murky underworld of altcoins.

The SEC in late March asked the U.S. District Court for the Southern District Court of New York to order Ripple to comply with federal securities laws and pay a $1.9 billion fine to the SEC. According to the court, Ripple is expected to file an opposition brief on April 22 and the SEC would reply on May 6. The court says it will issue a final ruling on remedies sometime after that. 

Ripple declined a request for comment, but pointed to a recent tweet posted by CEO Brad Garlinghouse: "Gensler's SEC has repeatedly acted outside the law — not going unnoticed by Judges admonishing the agency for a 'gross abuse of the power entrusted to it by Congress' (DEBT Box case) and for acting without 'faithful allegiance to the law' (Ripple case). Let's not also forget Gensler's lack of attention to SBFraud."

Click here to read the full story.
Mature woman feels confused experiencing difficulties with laptop usage
Banks reported nearly $27 billion had been tied up in scams or theft against elderly people in a recent 12-month period, according to a report from the U.S. Treasury.
fizkes - stock.adobe.com

Strangers, family defrauding seniors at an alarming rate: Fincen

Article by Carter Pape
A bureau of the Department of the Treasury has released an analysis calling attention to a scourge of financial exploitation against elderly people after it received more than 155,000 suspicious activity reports on the matter over a one-year period.

The analysis from the Financial Crimes Enforcement Network, or Fincen, follows up on a June 2022 advisory on what it calls elder financial exploitation, which is the illegal or improper use of the funds, property or assets of someone 60 years of age or older. In the year following that advisory, financial services companies (mainly banks, according to Fincen) reported $27 billion had been tied up in actual or attempted transfers relating to elder financial exploitation.

Elder financial exploitation encompasses both elder scams, which is the transfer of money to a stranger or imposter for a promised, undelivered benefit; and elder theft, which is theft by an otherwise trusted person. According to Fincen's analysis, roughly 80% of exploitation reports involved scams, and the remaining 20% were theft cases.

Click here to read the full story.
Sankaet Pathak, Founder and CEO of SynapseFI.
Sankaet Pathak is the founder of Synapse, which filed for bankruptcy.

What Synapse's bankruptcy means for the BaaS model

Article by Miriam Cross
A banking-as-a-service middleware platform's slide into bankruptcy is another sign of trouble for this business model, where software providers connect licensed banks and nonbank companies that want to take deposits and extend loans — and sometimes promise to offload part of the compliance burden

It also suggests the remaining players will have to evolve to survive. 

On April 19, instant payment platform TabaPay announced that it had entered into an agreement to acquire the assets and affiliates of Synapse Financial Technologies after Synapse filed for chapter 11 bankruptcy, pending approval in the bankruptcy court. Synapse, which supported banking-as-a-service relationships by connecting nonbank companies aiming to launch or embed financial products with licensed banks, has been beset by struggles over the past year, with two sets of layoffs, the loss of a major client, Mercury, and a clash with a partner bank over money. 

Click here to read the full story.
Fed Pivot To Rate Cuts Raises Optimism For Quicker IPO Revival
Despite advancements in AI for transaction monitoring, financial institutions share little in the way of fraud data, undermining efforts to combat crimes including check fraud, according to a recent report from the U.S. Department of the Treasury.
Valerie Plesch/Bloomberg

Treasury accuses banks of 'insufficient data sharing' on fraud

Article by Carter Pape
In a report last month on AI and cybersecurity, the U.S. Department of the Treasury said that, while banks tend to share plenty of information with each other for the purposes of cybersecurity and anti-money-laundering, they have practiced "insufficient data sharing" in the area of fraud prevention.

The dearth of banks sharing their fraud data undercuts smaller banks' efforts to train anti-fraud AI models  models that many banks hope will replace rule-based engines, deny lists and device fingerprinting in the fight to detect and prevent transaction-related crimes such as money laundering and fraud.

Treasury acknowledged a general gap in the data available to financial institutions for training AI models of all kinds, but the report said the gap is "significant in the area of fraud prevention," which the report contrasted with robust cybersecurity data sharing efforts led by organizations including the Financial Services Information Sharing and Analysis Center.

Click here to read the full story.
February 20, 2018 Palo Alto / CA / USA - Entrance to the Main Qu
Tests of models from OpenAI, Google and others found their advice tends to disadvantage names commonly associated with racial minorities and women.
Sundry Photography - stock.adobe

Stanford researchers say large language models favor white-sounding names

Article by Penny Crosman
Popular generative AI models like OpenAI's ChatGPT respond differently to prompts involving Black- and white-sounding names, according to tests conducted at Stanford University's Human-Centered Artificial Intelligence center.

In a paper entitled, "What's in a Name? Auditing Large Language Models for Race and Gender Bias," Stanford law school research fellow Alejandro Salinas de Leon and professor and associate director of Stanford's Human-Centered AI center Julian Nyarko asked several large language models for advice, using different names. Names commonly associated with white men, such as Dustin, Hunter and Jake, produced the most favorable results. Names associated with Black women, such as Keyana, Lakisha and Latonya, received the least advantageous outcomes. 

For those in the financial industry that have begun using large language models to draft emailsto detect fraudto summarize customer calls in contact centers, to rewrite core systemsto help financial advisors advise their clients and for other purposes, this confirms the worry that in addition to the dangers of errors, hallucinations, data privacy violations and copyright infringement, large language models could unintentionally introduce bias and impact the decisions of the people using them. 

Click here to read the full story.
New York, USA - 15 February 2021: American Bankers Association ABA website in browser with company logo, Illustrative Editorial.
The ABA is testing an information-exchange network to allow banks to share their fraud data with one another. The network could launch as a pilot sometime this fall.
Adobe Stock

Sharing fraud data is hard. Two industry efforts aim to make it easier.

Article by Carter Pape
As banks fight to reduce rampant fraud, regulators such as the U.S. Department of Treasury have suggested that the industry needs more collaboration to catch fraudsters. But as Europe finalizes regulations that would enable just this kind of fraud data sharing, no such rules exist in the U.S.

Because of this lack of collaboration, fraudsters benefit from the fact that even if they get caught defrauding one bank, it might take their next target days or weeks to learn about it. This gives them time to carry out their next scheme.

Multiple efforts are underway from fraud detection companies, banking industry groups and others to establish data-sharing exchanges for banks looking to fight back. Among these efforts, and cited by the Treasury in its recent report touching on the fraud data ecosystem, is the American Bankers Association's fraud data exchange.

Click here to read the full story.
Vancouver, CANADA - Feb 1 2023 : Website of ChatGPT seen in an iPhone. ChatGPT is a chatbot by OpenAI and is built on top of OpenAI's GPT-3 family of large language models.
A new, free leaderboard created and continuously updated by the Kensho team at S&P Global gives people in the financial industry a sense of which generative AI models work best for specific use cases.
Koshiro - stock.adobe.com

Which large language models are best for banks?

Article by Penny Crosman
As banks experiment with large language models like OpenAI's GPT-4 and Anthropic's Claude — artificial intelligence algorithms trained on large datasets to understand, summarize, generate and predict new content — a trend has emerged: In this area as in many others, some are leaning heavily on large cloud providers like Google, Microsoft and Amazon.

"Most banks — excluding perhaps the biggest tier-one banks with substantial research and development in-house — will be much more focused on the technology providers they work with to build these capabilities, in other words the IBMs, Microsofts, Googles, Oracle and AWSs of the world, rather than the much more hyped-up foundation model providers like OpenAI," said Gilles Ubaghs, a strategic advisor at Datos Insights.

But this could be a mistake, according to Bhavesh Dayalji, chief AI officer for S&P Global and CEO of Kensho, an AI company S&P Global bought in 2018. There is risk for companies in locking themselves into a closed architecture and one model or type of model from a big cloud provider, Dayalji said.  

Click here to read the full story.
Amex bolsters digital design to boost customer satisfaction.jpg
“The more we added to our business, the more we added to our digital channels, so we had to rethink the way those channels were structured,” said Evan English, at left, who leads Enterprise Design & Research at American Express. Ryan Cailliet, vice-president of product design and research is at center; Danny Forst, vice-president of product design, is at right.

Amex bolsters digital design to boost customer satisfaction

Article by Miriam Cross
American Express has been shifting and melding its design teams for several years, in a quest to improve the digital customer experience as a whole. In 2023, Amex formed Enterprise Design and Research, which brought membership and acquisition design teams under one roof. 

At 150 people, it is the largest digital design organization within American Express, although there are smaller teams embedded in other parts of the company, including the commercial and servicing businesses.

Digital design is about building functionality that helps users understand what they need to without a direct conversation. Customer satisfaction, especially in digital, has become paramount at Amex.

Click here to read the full story.
Discover deploys generative AI in its contact center.jpg
Discover is using generative AI to help agents quickly answer customer questions, rather than having to sort through policies that are hundreds of pages long, said Szabolcs Paldy, senior vice-president of operations at Discover, pictured at left. Yolande Piazza, vice-president of financial services at Google Cloud, is at right.

Discover deploys generative AI in its contact center

Article by Miriam Cross
Discover Financial Services is using generative artificial intelligence from Google to simplify the way agents research and address customer queries in its contact center.

On April 9, the two companies announced that the $152 billion-asset bank is deploying Google Cloud's AI development platform, Vertex AI, to summarize policies and procedures for agents so they can quickly return answers to customers while on the phone, and enable real-time search for documents using natural language. 

One major goal is to alleviate the mental load on agents so they can better focus on the customer. As digital banking has progressed, "The calls that customer service agents are getting today are the most difficult calls," Szabolcs Paldy, senior vice-president of operations at Riverwoods, Illinois-based Discover, said in an interview. "You don't call for your balance or to sign up for an offer."

Click here to read the full story.
MORE FROM AMERICAN BANKER