The state of AI in payments in 2025: From generative to agentic and beyond

Artificial intelligence has seen growing adoption over the last few years, and is now having a significant impact in some parts of cross-border payments. We look at the state of AI usage by the payments industry in 2025.

Artificial intelligence (AI) is transforming the world of business across almost every vertical, and cross-border payments is no exception. While in 2024 the industry was seized by enthusiastic discussion of the potential of the technology, particularly the use of generative AI (genAI) and large language models (LLMs), 2025 is giving us an industry where AI is beginning to have a direct impact on many payments players. 

However, that impact is not evenly distributed. While some players are enthusiastically embracing the technology and incorporating it into public-facing products and services, others are far quieter about AI and its use. 

Given the direct operational potential of AI, there are naturally going to be applications that organisations opt not to discuss publicly unless they are looking to attract particular interest from investors. However, what is said publicly does give us a strong sense of who is doing what in the space – and where results are beginning to be felt.

In this report we conduct an exhaustive review of the public communications of cross-border payments companies, considering both discussions in earnings and the contents of press releases, to determine which companies are prioritising AI and which areas they are directing focus. 

Topics covered:

• How much is AI discussed in earnings calls?
• Cross-border payments’ discussion of AI: Analysing press releases
• Payments’ most vocal users of artificial intelligence
• Which areas of AI is the payments industry most focused on?
• Fraud, compliance and AML: Industry standards for AI or areas of new potential?
• Artificial intelligence’s potential for transaction optimisation
• Customer service in AI: From chatbots to onboarding
• AI as a marketing tool
• Agentic AI: Payments’ agent-focused emerging 2025 trend
• AI’s ongoing payments potential

How much is AI discussed in earnings calls?

While focusing only on earnings calls provides a relatively limited view of AI’s use in cross-border payments, our analysis of the frequency of mentions does lay some groundwork in terms of establishing how the industry has increased its focus on the technology. 

We looked at the earnings calls of 10 publicly traded companies that we cover regularly to determine which of them had discussed AI and how frequently it was mentioned. The companies in question were Corpay, Euronet, Flywire, Intermex, Mastercard, Payoneer, PayPal, Remitly, Visa and Western Union. 

We found that between Q1 2020 and Q4 2024, every company except Intermex had mentioned AI on an earnings call at least once, but that the amount it was discussed varied significantly, with only seven of the 10 mentioning the technology during an earnings call for financial year 2024 – Corpay, Intermex and Western Union did not.

Although companies’ discussion of AI is likely to have included or focused on generative AI much of the time, particularly over the last year, direct usage of the term was lower. Only three companies – Visa, Remitly and Mastercard – mentioned genAI directly in 2024 earnings calls, down from six in 2023, which also included Western Union, PayPal and Payoneer.

Crucially, however, the advent of genAI has coincided with a sharp uplift in broader discussions of AI in earnings calls, with the collected mentions of artificial intelligence across all of the 10 companies’ call transcripts in financial year 2020 totalling just 10, but jumping to 78 in 2023 and reaching 100 in FY 2024. 

This reflects the impact that the launch of generative AI products such as ChatGPT has had on the industry, and on expectations around the use of AI among such businesses. However, the precise nature of what companies are using the technology for varies.

Cross-border payments’ discussion of AI: Analysing press releases

To get a richer understanding of how the cross-border payments industry is using AI at present, we turned our attention to the press releases of companies in the space. This saw us analyse the PR output of 61 of the companies on The 2024 Cross-Border Payments 100, our market map of the top 100 companies in the industry. 

Here we reviewed the majority of companies in the private equity-backed, VC-backed, public companies and independently owned sections, excluding a small number with major businesses outside the payments space. We also opted not to include companies in the ecommerce, mobile and bank categories as these companies typically have a far broader remit that means a greater number of their press releases will not have relevance to the cross-border payments industry. 

We analysed every press release published by these 61 companies between 1 January 2024 and 13 March 2025 – a total of 1,173, although three of the players released no press releases at all during this time. We found that AI was mentioned in 124 of these press releases (11% of the total), with 35 press releases (3%) where it was a primary or major topic of focus. 

However, there was some variation by the type of company the release was from. Among B2B2X providers – companies that specialise in providing payments networks and other infrastructure to other companies – a greater share of PRs mentioned AI than for any other industry, with 15% of all PRs issued by companies in this sector mentioning AI. In second place was payment processors, with 13% of PRs issued by companies in the sector mentioning AI, while B2B payments and consumer money transfers had shares of 10% and 4% respectively.

Crucially, this is not simply a case of a small number of companies putting out large numbers of press releases on AI. Out of the 58 who did release at least one PR during the time period reviewed, 33 (57%) mentioned AI at least once, while 14 (24%) issued a PR where AI was the main focus.

Again looking at the assessed companies by sector, payments processors were the most prolific in their discussion of artificial intelligence, with 88% of the payments processors reviewed issuing at least one PR mentioning AI, closely followed by B2B2X (85%), with B2B payments and consumer money transfers in third and fourth at 43% and 36%.

While rates vary widely, the fact that even consumer money transfers – a sector that includes some highly retail-based players – saw AI mentions from more than a third of companies in the sample speaks to the prevalence of the technology across the industry.

Payments’ most vocal users of artificial intelligence

While many companies discussed AI in press releases at some point over the past 15 months, there were a relatively small number who mentioned the technology in notably greater numbers than others. In terms of the number of press releases mentioning the technology, Mastercard was the clear leader, issuing 23 press releases in which AI was mentioned in some form since the start of 2024, 11 of which had AI as a primary or sole focus. 

This represented a significant step up from the next most frequent discusser of AI, PayPal, which issued 13 press releases mentioning the technology, four of which had it as the main focus. Papaya Global came in third while Stripe followed in fourth, although it had the third-most press releases where AI was the primary focus.

However, when considered as a share of the overall number of press releases the company released during the period, the picture was slightly different, as there was significant variation between the number of press releases different companies released. While 25 of the 58 companies released less than 10 press releases during the period, Mastercard released 83, the most of the group, followed by PayPal’s 63. 

Taken as a share of the PRs that each company released, global workforce payments players Papaya Global and Deel mentioned AI most frequently, with the technology featuring in 86% and 80% of their PRs respectively. However, when looking specifically at PRs where AI was the main focus, Swift had the greatest share at 22%, followed by Deel’s 20% and Stripe’s 15%.

Which areas of AI is the payments industry most focused on?

While many companies discussed various aspects of AI in the press releases they published, there were some areas that saw more widespread discussion than others.

General AI discussion, which typically saw companies mention the abstract potential of the technology or refer to events or businesses focused on it, was the most commonplace area of focus in the PRs analysed for this report. 42% of the 124 PRs mentioning the technology had discussion that fell under this category.

Beyond this, discussion of AI to combat fraud or support know your customer (KYC) or anti-money laundering (AML) compliance requirements were the most common application focus, with 28% of the AI-mentioning PRs focusing on this area.

AI solutions to optimise transactions or treasury management, or otherwise combat transaction failures, came in third with 15%, with solutions related to customer service or onboarding in fourth (10%) and marketing, customer retention and corporate and social responsibility (CSR) related initiatives in fifth (8%).

This was broadly reflected in the AI-focused press releases of key players. Most discussed fraud or related topics, while a significant number also discussed AI for transaction optimisation or customer services, with fewer focusing on its applications for marketing.

Notably, none of the players we reviewed mentioned applications relating to all four major topics (fraud, customer service, marketing and transaction optimisation). However, many of the companies that were more prolific in issuing press releases about AI also discussed a wider range of applications for the technology, with Stripe, PayPal and Mastercard all discussing AI applications in three out of the four areas.

Fraud, compliance and AML: Industry standards for AI or areas of new potential?

While AI to combat fraud and support compliance initiatives is a focus for companies across the spectrum of cross-border payments, it is more commonplace for some than others. 

Of the companies that issued at least one PR mentioning AI from 1 January 2024 to 13 March 2025, 71% of payment processors discussed fraud and compliance-related applications. This was followed by 64% of B2B2X players, 40% of consumer money transfers providers and 20% of B2B payments providers.

Prior to the advent of generative AI, earlier iterations of artificial intelligence had already found widespread use in the payments industry for specific applications, many of which were focused on compliance and security-related applications. AI has for years, and in some cases decades, been used to support anti-fraud efforts and streamline AML and KYC processes.

As a result, it has often been assumed that discussion of AI in fraud and compliance settings is merely reiterating established AI practices in an attempt to jump on the trend for increased AI discussion, rather than a true reflection of innovation with the technology. However, our analysis of press releases in this area shows this has not been the case with announcements over the past year or so. 

Of the payments companies who most prolifically discuss the use of AI to combat fraud and support compliance in press releases, many explicitly refer to the use of generative AI or related natural language processing technologies for this application.

Mastercard, which led mentions of fraud-related AI solutions, announced a genAI upgrade to its Decision Intelligence platform in 2024 that is designed to review data points surrounding a transaction, including key real-time information, to predict the likelihood of it being genuine. Stripe, meanwhile, launched a solution targeting online fraud that enables businesses to input new fraud rules using natural language prompts.

Even where genAI is not specifically referenced, companies have often launched new products or features that focus on using AI for fraud applications. For example, at the end of 2024, following a multi-institution pilot, Swift launched an AI-powered anomaly detection service to support banks in combatting financial crime.

Artificial intelligence’s potential for transaction optimisation

Away from fraud, there was also a noticeable trend for discussion of the use of AI to optimise transactions and so combat inefficiencies in the process that lead to delays or failures. 

This was most common among B2B2X players, appearing as a topic in 55% of PRs issued by such companies that mentioned AI, followed by B2B payments (30%) and payment processors (29%) but notably absent among consumer money transfers players.

Such optimisation typically involves AI determining optimum routes or options for a specific transaction, although it varies in nature depending on the sector or application. 

Stripe, for example, outlined its Optimized Checkout Suite, which determines which payment methods to offer a specific customer, while Checkout.com was among those to highlight the use of AI-based dynamic routing to streamline transaction processes. Thunes also shared its use of AI to power an in-house smart treasury platform.

Customer service in AI: From chatbots to onboarding

While fraud and payment optimisation are largely under-the-hood technologies, the use of AI in customer service is generally more directly clear to clients and users, and is often the most direct way they will experience a company’s use of the technology. Adoption of it for this purpose is therefore a potential risk if it is not sufficiently mature, and also serves as a vote of confidence in the technology and what it can do.

Here, there was a very sharp divide between industries that typically cater to high-volume consumer customers (consumer money transfers and payment processors) and those that cater to smaller numbers of business customers. While B2B2X saw this application mentioned in under 10% of PRs that discussed AI use, and B2B payments saw no mentions at all, customer service applications were mentioned in 40% of those for consumer money transfers and 43% of payment processors’.

Mastercard is the leader in this space in terms of press releases mentioning AI applications related to customer service, though much of its discussion in this area focused on solutions developed with partners, such as an AI-powered shopping assistant it produced for luxury brand Michael Kors.

However, in terms of AI alternatives to customer support, digital remittance player Remitly is the most prolific. The company has made extensive investments into AI for customer support, as highlighted in our recent discussion with CEO Matt Oppenheimer, and in November issued a release saying that its virtual support assistant had reduced the average time a customer spends with support by 75%, with only 3% of those triaged by the system opting to speak to a human representative.

AI as a marketing tool

The use of AI to support marketing and customer retention, meanwhile, is less widespread, but remains a significant presence across major verticals. 29% of AI-discussing PRs from payment processors, 20% of consumer money transfers providers, 18% of B2B2X players and 10% of B2B payments companies mention the application.

Notably, however, marketing and related fields are an area that companies are less likely to openly discuss their use of AI, particularly given that press releases are inherently a tool of marketing. It is quite likely that at least a small minority of the press releases we analysed were written either entirely or in part using AI tools such as OpenAI’s ChatGPT or Google’s Gemini, however current technology cannot say with sufficient certainty whether this is the case or not.

Despite this, there are companies who have incorporated AI into their marketing and related efforts. Mastercard, for example, was among those that focused on AI for some of its CSR efforts, including through the Artificial Intelligence to Accelerate Inclusion Challenge, a global initiative that awards projects using AI to improve financial inclusion. 

Meanwhile MoneyGram had one of the most overt uses of AI in its marketing efforts with its “Make Your Mother’s Day” campaign. Designed to tie into Mother’s Day in the US, this saw the company encourage users to make digital cards using a dedicated generative AI platform, which they could then share online to be in with a chance to win a trip home.

Agentic AI: Payments’ agent-focused emerging 2025 trend

While much of the focus of AI discussion over the past year or so has focused on relatively established areas for the technology, the last few months has also seen the quiet emergence of what appears to be a strong potential growth area for artificial intelligence: agentic AI. 

This is the use of AI agents to autonomously execute processes, including transactions, without ongoing human involvement, and it is an area that many in the payments space see significant future potential in. A Citi report on the potential of the space published in January said that agentic AI “could have a bigger impact on the economy and finance than the internet era”, while other discussions have similarly characterised it as transformational. 

While there were almost no discussions of agentic AI in the press releases we analysed, with Stripe providing a minor exception, discussion from broader payments-related players indicates this is an area with strong potential. As part of their yearly letter, Stripe’s CEO and President John and Patrick Collison outlined how the company had recently launched some solutions to enable virtual AI agents to make payments more smoothly, rather than having to interface with a human-designed checkout, as well as spend money on dedicated virtual cards. 

Meanwhile, other players including eBay, PayPal and India’s Razorpay have begun to explore the use of AI agents to make payments, while Intuit is exploring the technology to process tasks related to accounts payable and receivable.

The ability to accept agentic payments appears to be an area of particular potential, if investments in startups are anything to go by. Several recently founded startups have attracted investment for their focus on enabling agentic AI payments, including Skywire in the US and Nevermined in Switzerland. Meanwhile, there are also some players developing solutions to bring agentic AI to banking.

AI’s ongoing payments potential

Artificial intelligence attracted immense hype across 2024, but within that we are beginning to see genuine impacts to products, solutions and features, many of which have only been launched recently and which are likely to continue to see iteration and development over 2025 and into the next few years. 

While it is clear that some impacts for the industry are more powerful than others, certain players are beginning to strongly differentiate themselves by how much they focus on AI – and some are starting to see a measurable return on their efforts. 

The steady emergence of agentic AI indicates that generative AI is far from the peak of the technology, but one step on an evolving path for the industry and the wider business world beyond it. However, it is also clear that AI is not yet proving its worth for everybody, and there are still many players – some of them big names in the industry – who remain far quieter about the use of the technology in their businesses.

AI has many potential use cases, and some will better suit some players than others, but it is clear that much of its potential is still being explored and, barring a major shock, its use in payments is likely to only grow over the next few years.