<div id="readability-page-1">Large payment companies are speeding up agentic AI projects, putting pressure on banks to keep up.PayPal and Google plan to collaborate on agentic commerce, a form of artificial intelligence that requires little or no human supervision to perform tasks such as shopping and checkout. PayPal and Google's partnership quickly followed Mastercard's announced release of a partnership with Stripe, Google and Ant International subsidiary Atom to scale agentic payments; and Google's announcement of an agentic AI protocol. &#34;All banks will need to look at how they support this emerging channel for payments,&#34; Zachary Aron, principal at Deloitte Deloitte Consulting LLP, told American Banker. <h2>Big payment moves</h2>PayPal and Google's partnership covers several areas, including agentic shopping and commerce, embedding PayPal technology in Google's platforms, expanded payment processing and collaborating on cloud-delivered technology.Google and PayPal will cooperate on AI-supported shopping experiences and standards for agentic AI that the two companies hope will be adopted by the broader payments industry. PayPal will contribute its payment processing systems, data-driven personalization and identity technology. Google will add its own AI tech, developed with Nvidia and other partners, and will rely on PayPal's scale to boost adoption of Google's Agent Payments Protocol. PayPal and Google did not comment. Google's agentic AI protocol supports debit, credit, stablecoin and real-time transfers involving AI agents. Google recruited more than 60 companies to write the protocol, including Adyen, Alipay, American Express, Mastercard, PayPal and Worldpay.In an unrelated project, Mastercard is also pushing agentic commerce and plans to dramatically scale availability of the technology. By November 28 (Black Friday), all Mastercard cardholders will be enabled for Mastercard Agent Pay in the U.S., with global deployment coming shortly after. Mastercard additionally released Agent Sign-Up, which lets the card network's Agent Toolkit users identify AI agents and access AI-enabled Mastercard products. Other products are Insight Tokens, which enable security for agentic commerce, and agentic consulting services. Citi and U.S. Bank are the first announced bank partners for Mastercard's expanded AI shopping tools, which will lead to new services for the banks' Mastercard cardholders. The announcements from PayPal, Google and Mastercard are about building the infrastructure for agentic commerce when it arrives, Aaron McPherson, principal at AFM Consulting LLC, told American Banker, noting there is not a lot of consumer demand for agentic commerce at this point, but there are &#34;numerous problems to be solved.&#34;This includes who is responsible when an agent buys the wrong thing, how merchants make their catalogs readable by agents and the possibility of an agent being hijacked by a malicious website and being instructed to do things that violate a user's intent or privacy. &#34;Remember, AI can't tell a sketchy website from a legitimate one, so there is a lot of opportunity for shenanigans,&#34; McPherson said. <h2>Big work for banks</h2>While using agentic AI is relatively new in financial services, it is poised to grow quickly. By 2030, agentic commerce could drive up to $17.5 trillion in commerce, according to Deloitte, which says agentic AI will be a &#34;pivotal force&#34; in the future of payments and consumer engagement. &#34;Commerce is the best example of how a bank is present in the everyday life of its customers through embedded payments,&#34; Nico Kaplun, CEO of Finance AI Studio at Globant, told American Banker. &#34;Like any new technology impacting the customer, banks must develop a clear roadmap for adopting agentic commerce.&#34;For banks, agentic AI can lower operational costs by 20%, McKinsey senior partners Pradip Patiath and Ido Segev wrote in American Banker. The McKinsey partners also said banks will need to upgrade their technology, operating models and fraud prevention, adding expense to technology budgets. &#34;If that's not enough, agentic AI could also cut into revenues and profit margins,&#34; Patiath and Segav wrote. &#34;For example, AI agents could autonomously open new savings accounts and move money on consumers' behalf to find them the best rates. Or agents could optimize credit card lending balances and take advantage of zero balance-transfer offers, threatening margins.&#34;As payments and technology companies adopt agentic AI, banks will need to determine a strategy based on developing their own solution, partnering or integrating their products with a solution provided by a merchant or non-bank, Deloitte's Aron said, adding considerations include the degree to which banks can capture wallet share for the transaction, and insight that can help them better serve their customers or strengthen their pitch to merchants. &#34;While we believe the technology will evolve rapidly and customer-facing solutions will be introduced quickly into the market, a big determinant of whether the solutions are widely adopted and scale will be determined by merchant adoption,&#34; Aron said. &#34;Solutions which make it easier for merchants to offer their goods and services through an agentic solution or create a personalized, contextual experience for the customer will drive greater adoption.&#34; <h2>Complicated processing</h2>Agentic commerce creates a number of challenges across the payment value chain, including banks, merchants, and consumers, Aaron Press, research director for worldwide payment strategies for IDC, told American Banker. &#34;Understanding the data that will be needed and the implications of this data is in the early stages, though it is encouraging that concepts like fraud and security are being considered from the beginning,&#34; Press said. &#34;The mechanisms and data management process for all of this, however, are in the early stages.&#34;Banks will need the ability to confirm that the agent is owned and controlled by the account holder, and that the account holder intended for the transaction to take place. &#34;The process will also look different depending on the payment type, with card-based payments requiring a different model than DDA-based [demand deposit account] payments,&#34; Press said. &#34;This becomes even more complex as you move around the world and try to incorporate the huge number of local and regional payment schemes.&#34;It's also likely that tokenized credentials and digital wallets will be critical to this process, according to Press. &#34;Banks would be well advised to understand the implications of token management anyway, but it will be more important in an agentic commerce world,&#34; Press said.</div>

万事达

PayPal、谷歌和万事达正在推动自主 AI 商业，促使银行迅速适应。PayPal 和谷歌正在合作开发基于 AI 的购物体验，而万事达则在扩大其自主商业技术的规模。尽管消费者需求有限，但预计自主 AI 将对商业产生重大影响，可能在 2030 年前达到 17.5 万亿美元。银行必须制定策略以整合这项技术，这可能降低运营成本，但也可能威胁到收入。支付的未来将取决于银行和支付公司在多大程度上能够适应这一新兴格局

PayPal, Google and Mastercard are all in on agentic AI

- 大型支付公司加速代理 AI 项目，迫使银行跟进。  
- PayPal 和 Google 合作开发代理商业，涉及多项支付技术。  
- 预计到 2030 年，代理商业将推动高达 17.5 万亿美元的交易。

PayPal、Google 和万事达卡都在积极投入代理型人工智能