Hello! Welcome to Skit.ai. Click here to book a demo.

How Is the U.S. Planning To Regulate AI in Financial Services?

From automation and decision-making to fraud detection and customer experience, the applications of artificial intelligence in financial services seem endless. As companies, both large and small, navigate this evolving landscape and its plethora of vendors and solutions, many ask themselves: How will this technology be regulated once our legislative branch starts looking into it more seriously?

At Skit.ai, we organized a panel discussion hosted by our friends at Accounts Recovery with three renowned experts, to whom we asked the most pressing questions on AI in financial services and the regulatory environment. What regulations should we expect? More specifically, which aspects of AI will regulators be more interested in scrutinizing?

In this article, we’ll discuss the current role of AI in the financial sector — with particular attention to the accounts and receivables industry — and report some of the insights from the industry experts we interviewed during the event.

Understanding AI’s Impact on Financial Services

AI in financial services is not a prediction or a catchphrase. According to an international survey published in 2020 by the World Economic Forum and the Cambridge Centre of Alternative Finance, 85% of financial services providers already use AI in some form. Additionally, 77% of the responding institutions reported believing that AI would become essential to their business in the following two years. With the launch of ChatGPT in 2022, these numbers can only be higher now.

Some of the most notable applications of AI in the sector, according to Deloitte, are:

  • Conversational AI (such as chatbots and voicebots) for consumer interactions
  • Fraud detection and prevention
  • Customer relationship management
  • Predictive analytics
  • Credit risk management

The Regulatory Framework in the United States

Over the last few years, there have been efforts for legislators to study and regulate the use of AI in various industries, including the financial services industry. But while other foreign legislative bodies have been notably faster than the U.S. at passing timely legislation, there has yet to be a successful attempt at the federal level here in the United States.

In 2022, a bipartisan privacy bill, the American Data Privacy Protection Act (ADPPA) was introduced in Congress, but it did not make it through the Senate and has ever since been abandoned. Later in 2022, the White House published a policy document named the “Blueprint for an AI Bill of Rights,” seeking to provide guidance on the different rights that lawmakers should keep in mind when framing the discussion on the regulation of AI across industries.

In September, the U.S. Senate Committee on Banking, Housing, and Urban Affairs held a hearing about “Artificial Intelligence in Financial Services” to discuss AI’s applications, risks, and benefits in the industry.

The witnesses who spoke at the hearing were Melissa Koide of FinRegLab, who spoke about credit underwriting; Professor Michael Wellman of the University of Michigan, who raised concerns about algorithmic trading and market manipulation; and Daniel Gorfine of Gattaca Horizons, who focused on the opportunities presented by AI.

Most recently, the White House issued an executive order on artificial intelligence, establishing guidelines for AI safety and security. The order includes requirements that aim to protect consumers from threats to privacy, discrimination, and fraud.

Insights from the Experts: Possible U.S. Regulations of AI

The following quotes are excerpts from the webinar hosted by Accounts Recovery. Watch the recording to listen to the entire conversation and get the full context. The four experts who spoke are Dara Tarkowski of Actuate Law, Heath Morgan of Martin Golden Lyons Watts Morgan, Vaishali Rao of Hinshaw Culbertson, and Prateek Gupta of Skit.ai.

(Please note: The information provided in this article does not, and is not intended to, constitute legal advice; instead, all information is for general informational purposes only.)

Key Takeaway 1: Look at the European Union for Guidance

The United States is pitifully far behind the EU, the UK, areas of APAC, and Australia in the way they’ve approached the technology and the utilization of the technology. If we want to see which direction our country will go in terms of AI regulations, we have a five-year playbook of what it looks like in the rest of the world.”

“What we’ve seen from the hearings that have been held in Congress; at its base, the concern by lawmakers and regulators and a lot of the practitioners, is that bad data leads to bad outcomes, which is selection bias. Then we’ve got process bias, which means that bad methods and bad processes lead to bad outcomes. Philosophically, those are the two issues that lawmakers are trying to address in whatever sector.”

“If you’re looking for guidance, put together a framework that is largely compliant with what the European Union has already laid out as the ethical and safe use of AI. In a global economy, it would be foolish of the United States to deviate too much from what the rest of the world is already adopting.”

Key Takeaway 2: This Is Not About Replacing People with Technology

“In our industry, the usage of these types of technologies is not and should not be to replace people or to replace the thoughtfulness and the consideration of the decisioning. However, a lot of these technologies can help speed up and improve our decisioning, so that people can make better and faster decisions, which is better for both businesses and  consumers.”

Key Takeaway 3: AI Must Provide Value to Consumers

When it comes to the use of chatbots and voicebots, “you can’t keep consumers in an infinite loop with the artificial intelligence system and not let them talk to an actual human being whenever the AI is unable to provide a resolution. One of the focuses needs to make sure that AI provides value to the consumer, and is not used as a way for companies to create a hurdle between consumers and live agents.”

Key Takeaway 4: Waiting for Regulations May Not Be the Best Strategy

Should we wait for regulations before adopting AI solutions to avoid any risks? “You can’t bury your head in the sand and say: ‘We’re not going to deploy this technology until there are regulations.’ It really isn’t a question of whether you are going to adopt this technology—it’s a matter of when. The more you accept that and look into having risk assessments, an AI policy, and an AI committee, the better you’re going to be. The technology is coming to you through vendors and consumers before you know it.”

Key Takeaway 5: Set up an AI Task Force

“Set up an AI task force, so you can set up a framework on how to use AI properly.”


Want to learn more about Conversational AI and how it can benefit your business? Use the chat tool below to schedule a free consultation with one of our experts!

How Skit.ai’s Voice AI for Debt Collections Complies with State-level Regulations

State-level Regulations Are Just as Important as the Federal Ones

Virtually everyone working in the accounts and receivables industry is familiar with Reg F, the law passed in 2021 to update the Fair Debt Collections Practices Act (FDCPA). Reg F provides parameters for call frequency in debt collections; in particular, the 7x7x7 rule, which allows a maximum of 7 calls in a 7-day period, and allows the collector to follow up only 7 days after having had a conversation with the consumer.

However, some states have stricter laws when it comes to the debt collection industry and call frequency.

When training new agents or deploying a new software solution for your collection strategy, it’s important not to forget these state-level regulations, which are just as important as the federal ones.

Examples of State-specific Regulations for Collection Calls

Here are three examples of state-level regulations that limit call frequency permissions further than Reg F.

Massachusetts: According to the Attorney General’s regulations, creditors and collection agencies are allowed to make a maximum of 2 attempts of communication via telephone (calls or text) in a 7 consecutive day period.

New York: New York’s law is similar to Massachusetts’. Also here, collectors are not allowed more than 2 attempts of communication (calls, texts, letters, emails, etc.) in a 7-day period.

North Carolina: Collection agencies are allowed to make only 1 attempt of communication to a particular third party in a 7-day consecutive period to obtain location information.

How Skit.ai’s Compliance Filters Tackle State Regulations

Working with legal and compliance experts, at Skit.ai we’ve compiled the different state-level regulations and have integrated them into our Voice AI solution’s compliance filters.

Our solution identifies the state of the consumer through the zip code of their most recent address and identifies the applicable regulations in real-time during the campaign initiation process. This way, Skit.ai’s solution never dials out a non-compliant call to a consumer.

Want to learn more about how Conversational AI can help you streamline your collection strategy and comply with all regulations? Schedule a call with one of our experts using the chat tool below.

Unpacking the TCPA for Debt Collection Calls with Voice AI

The debt collection industry is a heavily regulated space; the number of laws and regulations in place can be quite overwhelming. Whenever a new technology or solution emerges, therefore, it is natural to wonder whether it is compatible with the existing laws and whether the provider is fully compliant.

As more collection agencies look into adopting a Conversational Voice AI solution to automate their collection calls, it can be confusing to go through the regulations and determine which ones apply and which ones don’t.

In this article, we’ll unpack one important law — the Telephone Consumer Protection Act — and analyze its key provisions from the perspective of a Voice AI provider.

An Overview of the TCPA

The Telephone Consumer Protection Act (TCPA), first passed by the U.S. Congress back in 1991, is one of the most important laws that regulate telemarketing and the use of automated telephone equipment.

The TCPA is a law that governs and regulates all telemarketing calls, auto-dialed calls, pre-recorded calls, and unsolicited faxes. Though this law primarily focuses on protecting consumers from unwanted communications, it also governs and prescribes restrictions in the context of debt collection calls.

The TCPA authorizes the Federal Communications Committee (FCC) to exempt certain types of calls from its restrictions, including “calls made to residential lines that are not made for a commercial purpose, calls made for a commercial purpose that do not contain an unsolicited advertisement, calls from tax-exempt nonprofit organizations, and healthcare-related calls.”

In 2022, more than ​​1,500 TCPA complaints were filed in federal courts.

Does the TCPA apply to debt collection calls? And how does it affect the use of Voice AI?

The Key Provisions of the TCPA and Collections

Calling Curfew

Solicitors can’t call customers at night time (indicatively, between 9:00 p.m. and 8:00 a.m.). However, the specific hours are determined by each state. For example, certain states do not allow calls on Sundays (e.g. Alabama, Louisiana, and Mississippi, among others). During the week, the starting time when calls are allowed varies by state—between 8:00 and 10:00 a.m. Calls need to be interrupted between 6:00 and 9:00 p.m. depending on the state.

National Do Not Call List

The National Do Not Call Registry was created to stop unwanted sales calls; anyone can register their phone number. Good news! This provision only applies to telemarketing calls. Luckily, debt collections do not qualify as telemarketing. However, if the customer explicitly asks not to be called, the collection agency needs to honor the request.

Self-Identification via Voicemail

If the collection agency wants to leave a voice message to the customer, then the collector must identify themselves and the agency and provide their telephone number.

Calling Mobile Phones

Nowadays, fewer people have landlines at home, and virtually everyone owns a cellphone. Still, the TCPA rules that callers can’t call a mobile phone without prior consent when using an automatic dialer, artificial voice, or a pre-recorded message. Therefore, one must obtain direct consumer consent prior to the use of these technologies to contact cellphones. 

In the next section, we’ll see why Voice AI is not considered an automatic telephone dialing system (ATDS).

How the TCPA Impacts the Use of Voice AI

Over the years, there has been much confusion about what exactly qualifies as an automatic telephone dialing system (ATDS) under the TCPA. In 2021, the Supreme Court released its decision on Facebook v. Duguid, settling this long-standing uncertainty.

In a unanimous decision written by Justice Sonia Sotomayor, the Supreme Court established that an automatic telephone dialing system (ATDS) is a system or device that either:

  • Stores a telephone number using a random or sequential number generator;
  • Produces a telephone number using a random or sequential number generator.

Here’s why our Voice AI technology does not come under the purview of the ATDS definition under TCPA:

  • Skit.ai’s solution does not randomly or sequentially store or produce telephone numbers.
  • Our solution is designed to call the phone numbers provided by our clients, which are based on the accurate and complete databases of the clients, which enables  Skit.ai to perform the services throughout the contract period.
  • Since our clients have prior express consent to send communications to the identified consumers, Skit.ai can communicate with the consumers on behalf of its clients through its Voice AI technology.

Please note that the information in this article is not intended to be legal advice and may not be used as legal advice. For more information and to request a free demo, you can use the chat tool below to schedule a call with one of our collections experts.

Meeting Debt Collection Compliance With AI-Powered Digital Voice Agents

Owing to far-reaching repercussions, compliance management has become an issue of gravitas. It’s a challenge of change. Often, frequent regulatory changes create ambiguity for collection agencies. For instance, Regulation F of the Consumer Financial Protection Bureau (CFPB) came into effect on November 30, 2021, and is the most significant debt collection rulemaking. Any creditor–either the original issuer or a debt buyer–faces challenges in responding to it. And even more tedious is training and retraining agents, reiterative setting up processes and tools to meet regulatory requirements.

When it comes to compliance, the devil is in the details. A human agent under varying stress and performance pressure is prone to make mistakes. But even an innocuous breach of compliance results in hefty fines and penalties. Even without state or local mandates around debt collection practices, federal regulations must be followed to avoid penalties or lawsuits from consumers or enforcers. CFPB levied $1.7 billion in civil penalties and over $14.4 billion in relief for American consumers in the last ten years. Compliance has thus evolved as a significant pain point for debt collections agencies.



We have reached a point where compliance is not just an expense item but also a source of differentiation for collection agencies. Unsurprisingly, most debt collection agencies are looking for tech solutions that can help them be more agile and efficient. Voice AI is one emerging solution with the most disruptive potential and growing use cases.

Too Many Calls, Too Little Communication

One of the prime objectives of compliance is to protect the customer from unfair practices and harassment. CFPB bases much of its enforcement authority on the concept of UDAAP (unfair, deceptive, and abusive acts or practices).

A call at the right time, to the right person, and with the right message can achieve the 3 Cs of debt collection: Cost, Compliance, and Customer Experience. A human agent may struggle to accomplish the triad, making too many or too few calls, but it’s a cakewalk for an intelligent voice agent.

Explore how Voice AI solutions are Transforming Debt Collection

Current Compliance Challenges

The formal, statutory fees and levies, which are increasingly hefty, represent just the tip of the compliance cost iceberg (around 10%) of total regulatory costs. The broader cost of compliance is much bigger, making it a formidable force. 

Here are the common challenges faced by debt collection agencies today:

  • Ever-Expanding List of Laws: Fair Debt Collection Practices Act (FDCPA), Telephone Consumer Protection Act (TCPA), Federal Fair Credit Reporting Act (FCRA), Payment Card Industry compliance (PCI), and Health Insurance Portability and Accountability Act (HIPAA) are a part of a growing list of regulations, adherence to which is a core driver to the success of debt collection agencies and similar financial institutions.
  • High Cost of Continual Training and Vigilance Process: A survey of sector firms by the Credit Services Association (CSA) reveals that in staffing terms, the proportion of resources involved (in compliance) seems to trend generally between 15% and 25% of total resources. That is a significant percentage and an opportunity to cut down the cost.
  • Client Expectation and Audit Requirements: Clients of collections agencies are deeply wary of meeting compliance and exert pressure, even more than regulators, to comply. As per a report by CFPB, collection agencies with large clients face 17 audits in a year. That’s an average of 3 audits every 2 months. The lack of transparency between debt collectors and consumers makes it difficult for agencies to facilitate these audits effectively. It is a formidable challenge to meet such high expectations cost-effectively.
  • Insufficient Time to Design and Implement Compliance Effectively: A rapid and frequent change in regulation leads to collection agencies running from pillar to post to update their processes. Deploying AI-enabled voice agents can minimize the training and guidance cost.
  • High Cost of Not Meeting the Compliance Requirements: Failing to meet the compliance requirement has, in the past, led to grave heavy consequences. Encore and Portfolio Recovery Associates, two giants in bad debt collections, were fined $18 million in 2015. They were forced to refund or halt collection of over $160 million in consumer debts. Violating the Do Not Call registry can cost agencies anywhere between $500-$1500 per case, as per TCPA. Moreover, razor-thin margins make the total cost of attorney fees, settlement costs, and the opportunity cost of time too much for agencies to bear.

Voice AI and its Ability to Empower Collection Companies Manage Compliance

More often than not, compliance is a matter of adhering to protocols and procedures. AI-enabled digital voice agents that can religiously follow a given set of instructions prove far superior in adherence to the regulatory framework.

There are numerous instances where small mistakes land collection agencies in trouble. Here are some simple yet powerful examples of how Voice AI can help with compliances:

  • Honoring Do Not Call Registry and Data Scrubbing: The telephone Consumer Protection Act (TCPA) maintains a register of subscribers who do not want to be called for telemarketing calls and automated dialer calls unless you have consent to do so otherwise. It’s essential to scrub the data before dialing these contacts and check for permission. Solution is to scrub the data against certain database such as Do-not-call registries (external and internal), consumers represented by attorneys and debt settlement companies, deceased consumers, serial litigators, bankrupt consumers, cease-and-desist order consumers. Unlike human agents, who can fumble, digital voice agents perform this with the help of APIs in a fraction of a second.
  • Calling Within Permissible Hours: FDCPA does not allow collection agencies to contact customers outside of 8:00 a.m. to 9:00 p.m. local time unless the consumer has given explicit consent. Additionally, customers with night jobs may not wish to be contacted during the day. Such personalization in large portfolios prove to be a daunting task for a human agent but an effortless one for a digital voice agent.
  • Calling Frequency: Regulation F of CFPB limits the frequency of calls under the 7/7/7 rule, restricting the agencies from attempting to establish communication with their consumers for more than 7 times in 7 days. The 7/7/7 rule includes voicemail, unanswered calls, and messages left on the consumer’s phone, and excludes email and text messaging. Furthermore, agencies cannot try to establish contact in the next 7 days after a successful communication. It’s taxing for human agents to consistently follow these rules for the entire customer base while optimizing time and cost at the same time. On the other hand, configuring machines to follow all these rules is possible with a click. 
  • Mini-Miranda is mandatory as per FDCPA in the first communication in any channel. Digital voice agents never fail to comply with such regulatory requirements.
  • Failure to Discontinue Communication Upon Request: Communicating with consumers in any way (other than litigation) after receiving notice with certain exceptions can lead to lawsuits. Machines follow strict protocols and comply with the request submitted by the consumers.
  • Communicating with Consumers at Their Place of Employment: It’s illegal to contact the consumer after being advised that this is unacceptable or prohibited by the employer. Human agents under dier conditions fail to honor guidelines. On the other hand, since machines reachout at the right time and frequency have high conversion rate while meeting compliance.
  • Contacting a consumer represented by an attorney: Agents must not contact the consumers who have chosen not to be contacted by agencies and have signed up attorneys for communication with certain exceptions.
  • Communicating with a Consumer During Validation Period: Human agents can make a mistake and try to establish communication with the consumer or pursue collection efforts after receiving a request for verification of a debt made within the 30-day validation period. On the other hand, Digital Voice Agents are configured to not engage in any such activities and trigger the automatic collection calls once validation period is over.
  • Misrepresentation & Threatening Arrest or Legal Action: With variable incentive as a major wage component, it’s quite common for debt collectors to misrepresent as attorney or law enforcement officer. FDCPA prevents such kind of misrepresentation and has punitive enforcement directives. Digital voice agents follow strict protocol and never succumb to such malpractices.
  • The abusive or Profane Language used during communication related to the debt is prohibited. Digital voice agents never fall back to such practices in order to achieve the results.
  • Communication with Third Parties: revealing or discussing the nature of debts with third parties (other than the spouse or attorney) is prohibited except to know the location of the debtor without mentioning debt related information. Intelligent Voice Agents can confirm the right party before giving out any information.
  • Raise a Dispute: Voicebot can also help consumers raise a dispute over a call and tag it in the CRM so that the relevant team can pick it up.
  • Validation: Upon asking for validation information, the voice bot can immediately send the electronic copy of the validation notice and mark the contact with a relevant tag so that human agents can see the status, and neither the voicebot nor human agents try to communicate to the consumer for the next 30 days.
  • Raise Tickets: Voicebot can even raise tickets to send the physical copies of the validation notice if explicitly requested by the consumer.

With Distinct Advantages, Voice AI Will Play a Bigger Role in Compliance Management 

Apart from numerous other use cases, the utility of Intelligent voice agents in improving the compliance of debt collections agencies is fast emerging and very promising. 

Apart from the direct costs of compliance, indirect costs such as fines and penalties take a heavy toll on companies. Today, compliance has become more than an expense but a source of differentiation. Many companies have already begun adopting Voice AI, and its ever-expanding use cases will help them create a distinct competitive advantage.

For more information and free consultation, let’s connect over a quick call; Book Now!

Also, for more information visit our Collections Page.