Verified
Mastercard
Piloting- Payments network
- 39,800 employees
- AI owner: Greg Ulrich
- 80%+ of staff use AI (Oct 2026)
- No AI return reported
Mastercard builds AI into the services it sells: its models score transactions for fraud, and Agent Pay lets AI agents pay on its network. Inside the company, Mastercard runs AI through an agent platform that Chief AI and Data Officer Greg Ulrich calls an agentic factory. Ulrich said in Oct 2026 that more than 80% of Mastercard's employees use AI regularly. Mastercard has published no revenue, cost or headcount figure that it credits to AI.
mastercard.comTheir approach to AI
- The platform
- Mastercard builds its own AI, including fraud models such as Decision Intelligence and a large tabular foundation model trained on billions of transactions (Mar 2026). Mastercard's internal agentic factory has compliance, observability and guardrails built in (Jul 2026).
- The models
- Mastercard trains its own transaction models with Nvidia and Databricks, gives staff Microsoft 365 Copilot, GitHub Copilot and OpenAI tools, and has tested Anthropic's Claude Mythos and OpenAI's GPT-5.5-Cyber for security work.
- The reach
- Greg Ulrich said in Oct 2026 that more than 80% of Mastercard's employees use AI regularly. Microsoft 365 Copilot had reached 16,000 Mastercard employees by Sep 2025.
- The catch
- Mastercard has published no business return it credits to AI. Greg Ulrich says about a third of its services are predicated on AI, but Mastercard gives no AI revenue, cost or headcount figure. Mastercard tied its 2026 cut of about 4% of staff to a strategic review, not to AI.
Snapshot
- Employees
- 39,800
- Dec 31, 2025, about 70% outside the US (2025 Form 10-K)
- Technology budget
- Not disclosed
- Mastercard reports no technology or AI budget; operating expenses rose 12% in Q3 2026
- AI/ML headcount
- Not disclosed
- An AI Center of Excellence under Janet George (Feb 2025); 4,000 consultants use internal agents
- Who owns AI
- Greg Ulrich, Chief AI and Data Officer
- Since Apr 2024; Management Committee member; reports AI strategy to the board
- Platform
- Agentic factory, built in-house
- Internal agent operating system with guardrails and automated drift tracking (Jul 2026)
- Models
- Own models, OpenAI, Microsoft, Anthropic
- Large tabular foundation model with Nvidia and Databricks (Mar 2026)
- Reach
- 80%+ use AI regularly
- Oct 2026; Microsoft 365 Copilot for 16,000 staff in Sep 2025
- Use cases
- Fraud, agentic payments, consulting, coding
- No count published
Key findings
Mastercard has published no revenue, cost or headcount figure it credits to AI. Greg Ulrich said on Jul 14, 2026 that services make up about 40% of Mastercard, and that about a third of those services are predicated on AI and growing faster than the rest.
Greg Ulrich, Mastercard's Chief AI and Data Officer, told Forbes on Oct 21, 2026 that more than 80% of employees use AI regularly. Ulrich said AI Learning Hub sessions had drawn more than 13,500 attendances.
Mastercard built AI agents in 2025 for its 4,000 consultants, covering deep research, text-to-SQL, Excel and PowerPoint. Greg Ulrich said in Jul 2026 that, were Mastercard starting today, it would build them fundamentally differently.
Mastercard is training a large tabular foundation model on billions of anonymized transactions with Nvidia and Databricks (Mar 2026), with cybersecurity as its first target and the aim of maintaining fewer separate models.
Mastercard recorded a $202 million restructuring charge in Q1 2026 for actions affecting about 4% of its full-time employees, which it attributed to a strategic review of its business and to reinvestment, not to AI.
Where Mastercard sits
- Stage
- Piloting
Mastercard has reported no business return it credits to AI in the last 18 months, for any use case or for the company as a whole. Greg Ulrich said in Jul 2026 that about a third of Mastercard's services are predicated on AI and growing faster than the rest, but Mastercard publishes no revenue, cost or headcount figure for them. Mastercard's fraud model gains accrue to the banks that use them. On rollout, Ulrich, who has owned AI as Chief AI and Data Officer since Apr 2024 and sits on the Management Committee, told Forbes on Oct 21, 2026 that more than 80% of Mastercard's employees use AI regularly. Ed McLaughlin said on Nov 9, 2026 that all of Mastercard's developers have AI assistance, and that rollout places Mastercard at Piloting. Mastercard falls short of Scaling because it reports no AI-credited return in any function and no firm-level AI figure.
How the stages are defined - Ownership
- Chief AI and Data Officer Greg Ulrich, on the Management Committee
- Platform
- Hybrid: Own transaction models and agent platform; staff copilots bought
- Models
- Own fraud and foundation models, plus OpenAI, Microsoft and Anthropic
- Hosting
- Mastercard's own network data centers and a hybrid of public and private cloud; Microsoft's cloud for Copilot; providers for its own models not named
- Reach
- More than 80% of employees use AI regularly (Oct 2026)
How they deploy
AI inside the services it sells
Mastercard's 2025 Form 10-K names data and AI as one of the enablers of its strategy. The 10-K describes security, personalization, analytics and consulting services built on Mastercard's data and AI. Greg Ulrich said on Jul 14, 2026 that services make up about 40% of Mastercard, that about a third of those services are predicated on AI, and that the AI services grow faster than the rest. Michael Miebach's 2026 proxy letter cites more than 15 petabytes of structured data as the basis for that work.
Own models trained on own data
Mastercard trains its own fraud and risk models on its transaction data instead of relying on outside models for that work. Decision Intelligence Pro, announced in Feb 2024, applied generative AI techniques to real-time fraud scoring. In Mar 2026, Mastercard described a large tabular foundation model trained on billions of anonymized transactions. Distinguished Engineer Steve Flinter wrote that Mastercard maintains thousands of AI models for different markets, use cases and customers, a number the new model could cut significantly.
Agentic commerce as a network product
Mastercard launched Agent Pay in Apr 2025 so that registered AI agents can pay with tokenized card credentials. Mastercard enabled all US cardholders in 2025 and said on Apr 30, 2026 that nearly all Mastercards worldwide were enabled. Mastercard added Agent Pay for Machines for machine-to-machine payments in Jun 2026. Mastercard signed a broad strategic partnership with OpenAI in Q3 2026, which Michael Miebach said covers payments, joint services growth and AI transformation.
Internal agents built in-house
Mastercard builds its internal AI agents itself. George Maddaloni, then CTO of Operations, said in Sep 2025 that Mastercard had found success building internal agents on its own, with the Model Context Protocol opening up access to internal capabilities. Maddaloni said Mastercard was unlikely to use the protocol for customer-facing agents soon. Greg Ulrich said in Jul 2026 that Mastercard built an agentic factory, an operating system for agents with compliance, observability and guardrails built in.
Adoption through training
Mastercard requires training on its data policies before staff get access to generative AI tools, George Maddaloni said in Sep 2025. Mastercard also reviewed the workflows of each category of employee at the end of 2024 to see where the tools helped. Greg Ulrich said in Oct 2026 that a new AI Learning Hub offers role-based learning paths and that webinars, hackathons and office hours had drawn more than 13,500 attendances.
- Deployment model
- Mastercard runs AI from a central Data and AI organization under Chief AI and Data Officer Greg Ulrich. Mastercard created the organization in Apr 2024 to commercialize AI and data for internal and external use and to govern them across the company, with Chief Services Officer Craig Vosburg overseeing it. Janet George has led the AI Center of Excellence under Ulrich since Feb 2025. Mastercard Technology, under President and CTO Ed McLaughlin, runs the payments network, enterprise platforms and infrastructure. The Chief Security Officer, Michael Lashlee, leads vulnerability work with frontier models, and Chief Product Officer Jorn Lambert's Consumer Payments team owns Agent Pay. Mastercard has not said whether Ulrich's organization moved to Linda Kirkpatrick when she succeeded Vosburg as Chief Services Officer on Aug 3, 2026.
- Model providers
- Mastercard trains its own models for fraud, risk and personalization, including the large tabular foundation model it built with Nvidia and Databricks (Mar 2026). For staff, Mastercard rolled out Microsoft 365 Copilot to 16,000 employees and GitHub Copilot and other coding assistants to developers (Sep 2025). Michael Miebach said on Apr 30, 2026 that Mastercard uses OpenAI's tools as an enterprise customer. Greg Ulrich said in Jul 2026 that Mastercard was in the early wave of Anthropic's Project Glasswing with the Claude Mythos model and also worked with OpenAI's GPT-5.5-Cyber. Agent Pay launched in Apr 2025 with Microsoft and with IBM's watsonx Orchestrate for business payments. Mastercard planned to bring Microsoft's Azure OpenAI Service and Copilot Studio together with its payment tools.
- Data and infrastructure
- Mastercard processed more than 175 billion transactions in 2025 and holds more than 15 petabytes of structured data, according to its 2026 proxy statement. George Maddaloni said in Sep 2025 that Mastercard built a global hybrid cloud network so it can run public cloud, private cloud or both as each country's rules require. Mastercard has not named the public cloud providers that run its AI models. Mastercard built its knowledge agent program on the Databricks Data Intelligence Platform (Oct 2024). Mastercard trains its foundation model on Nvidia's accelerated computing platform with Databricks and removes personal data from transactions before training (Mar 2026).
Platforms and internal tools
Decision Intelligence and Decision Intelligence ProDecision Intelligence Pro announced Feb 2024
Decision Intelligence is Mastercard's real-time fraud scoring service for banks, which scored 143 billion transactions a year when Decision Intelligence Pro was announced in Feb 2024. Decision Intelligence Pro adds generative AI techniques that assess the relationships between the entities around a transaction in under 50 milliseconds. Greg Ulrich said in Jul 2026 that the newer technology finds 300% to 400% more fraudulent transactions in high-risk bands.
- Models
- Proprietary
Large tabular foundation modelMar 2026
Mastercard's large tabular foundation model is a generative model trained on billions of anonymized transactions. Mastercard plans to extend the model to hundreds of billions of transactions and to merchant, fraud, authorization, chargeback and loyalty data. Mastercard says the model beat standard machine learning techniques in early tests, for example on rare legitimate purchases. Mastercard is building APIs so teams across the company can build on the model. Cybersecurity is the first target.
- Models
- Proprietary, built with Nvidia and Databricks
Agentic factoryBy Jul 2026
The agentic factory is what Greg Ulrich called Mastercard's internal operating system for building agents, with compliance, observability and guardrails built in instead of added per agent. The agentic factory automates model drift tracking, which dedicated teams once did by hand (Jul 2026).
Consultant agents2025
Mastercard built a series of AI agents in 2025 for its 4,000 consultants, covering deep research, text-to-SQL, Excel and PowerPoint, because outside tools did not meet its needs at the time. Greg Ulrich said in Jul 2026 that, were Mastercard starting today, it would build them fundamentally differently.
- Users
- 4,000 consultants (Jul 2026)
Knowledge agent program and onboarding assistantOct 2024
Mastercard built a program for knowledge agent tools with Databricks. The first is a product onboarding assistant that answers customers' onboarding questions from Mastercard's documentation, using a large language model with retrieval and fine-tuning. Subject matter experts review the assistant's answers.
- Models
- Large language model, not named
Microsoft 365 Copilot and coding assistantsBy Sep 2025
Mastercard rolled out Microsoft 365 Copilot to 16,000 employees and gave developers GitHub Copilot and other coding assistants after a pilot with 2 small developer groups. George Maddaloni said in Sep 2025 that developers use the assistants for coding and writing tests, but that Mastercard had not yet adopted fully agentic coding, citing security and stability.
- Users
- 16,000 Microsoft 365 Copilot users (Sep 2025)
- Models
- Microsoft 365 Copilot, GitHub Copilot
Mastercard Agent PayApr 2025
Agent Pay is Mastercard's framework for payments made by registered AI agents, built on Agentic Tokens and its existing tokenization and dispute rules. Mastercard enabled all US cardholders in 2025 and said nearly all Mastercards worldwide were enabled by Apr 2026. In Q1 2026, Mastercard added Verifiable Intent, a tamper-resistant record of what a user authorized.
Agent Pay for MachinesJun 10, 2026
Agent Pay for Machines lets AI agents buy digital services such as APIs, compute and data in payments as small as fractions of a cent. Agent Pay for Machines handles credentialing, programmatic spending limits and settlement across cards, accounts and stablecoins. More than 30 companies, including Adyen, Coinbase, Cloudflare and Stripe, were among the first to support it.
Mastercard Agent SuiteAnnounced Jan 27, 2026
Agent Suite packages customizable AI agents and support from Mastercard's 4,000 advisers so banks and merchants can build, test and deploy agents, starting with product discovery for banks and conversational shopping for merchants. Mastercard said Agent Suite would be available in Q2 2026.
Vendors and partners
OpenAI2025
Enterprise AI tools for staff; Agent Pay partner; GPT-5.5-Cyber for security testing · Michael Miebach said on Apr 30, 2026 that Mastercard uses OpenAI's tools as an enterprise customer. Miebach said on Oct 29, 2026 that a new broad strategic partnership covers payments, joint services growth and AI transformation.
MicrosoftBy Apr 2025
Microsoft 365 Copilot and GitHub Copilot for staff; Agent Pay launch partner · Mastercard launched Agent Pay in Apr 2025 with plans to bring Microsoft's Azure OpenAI Service and Copilot Studio together with its payment tools.
Anthropic2026
Claude Mythos model through Project Glasswing, for finding software vulnerabilities · Greg Ulrich said in Jul 2026 that Mastercard was in Project Glasswing's early wave and that the models found vulnerabilities that had been hard to detect.
NvidiaBy Mar 2026
Accelerated computing for the large tabular foundation model
DatabricksBy Oct 2024
Data Intelligence Platform for knowledge agents and the foundation model
Google2026
Co-developed Verifiable Intent for agent payments · Michael Miebach said on Jul 30, 2026 that Verifiable Intent was developed together with Google.
IBMApr 2025
watsonx Orchestrate for business-to-business Agent Pay use cases
Recorded FutureDec 2024
Threat intelligence company acquired by Mastercard · Mastercard completed the $2.65 billion acquisition in Dec 2024 to add AI-driven threat intelligence to its cybersecurity, identity and fraud scoring services.
Leadership
AI, data and platform leaders
Executives
AI org map
Central Data and AI organization; AI sold through the services and product teams
Mastercard created a Data and AI organization in an Apr 2024 realignment to commercialize AI and data for internal and external use and govern them across the company. Chief AI and Data Officer Greg Ulrich leads the organization, and Chief Services Officer Craig Vosburg oversaw it at its creation. Ulrich sits on the Management Committee and briefs the full board on AI strategy. Mastercard Technology, under Ed McLaughlin, runs the network and infrastructure. Linda Kirkpatrick succeeded Vosburg as Chief Services Officer on Aug 3, 2026.
Mastercard's AI work sits in the central Data and AI organization under Greg Ulrich, which includes Janet George's AI Center of Excellence, and in Mastercard Technology under Ed McLaughlin. AI work also sits in the product and services teams that sell AI to banks and merchants, including some 4,000 consultants and advisers. Mastercard has not published how many data scientists or AI engineers it employs. Ulrich told Tech Brew in Aug 2024 that with billions of transactions a year it makes no sense to centralize data work in one department, so he coordinates data scientists across the company.
Reporting lines
President and CEO
Services
Linda Kirkpatrick, Chief Services Officer
Fraud, cybersecurity, identity, consulting, loyalty and open banking services; Craig Vosburg until Aug 3, 2026
Data and AI organization
Greg Ulrich, Chief AI and Data Officer
Overseen by the Chief Services Officer at its creation in Apr 2024; the current line is not published
AI Center of Excellence
Mastercard Technology
Ed McLaughlin, President and CTO
Payments network, enterprise platforms, infrastructure and information security
Consumer Payments
Jorn Lambert, Chief Product Officer
Agent Pay, stablecoin and core payments (Jun 2026)
Privacy, AI and Data Responsibility
Caroline Louveaux
Reporting line not published; briefs the board's Risk Committee
Mastercard's 2026 proxy statement lists Greg Ulrich and Michael Lashlee on the Management Committee without reporting lines. The 2026 proxy statement also says George Maddaloni, CTO of Operations, who led the rollout of internal AI tools, left Mastercard on May 29, 2026.
Where the AI workforce sits
Central Data and AI organization
Mastercard has published no AI headcount for the organization or the AI Center of Excellence.
Not disclosedServices, consulting and product teams
Estimate not possible: Mastercard says data scientists sit across the business, and its 4,000 consultants and advisers use AI agents but are not AI staff.
Not disclosed
Named people
Use cases
Mastercard's AI use cases fall into fraud and security services for banks and merchants, agentic payments on its network, personalization and consulting services, and internal tools for staff, consultants and developers. Mastercard has published no return it credits to AI for any of them. Mastercard's fraud figures measure detection gains for client banks, and Greg Ulrich's statement that about a third of its services are predicated on AI comes with no revenue figure. Mastercard reports no firm-level AI return.
| Use case | Area | Status | ROI |
|---|---|---|---|
| Real-time fraud scoring (Decision Intelligence) Mastercard scores each transaction on its network for fraud risk in under 100 milliseconds, on a scale of 0 to 999 passed to the issuing bank, across 175 billion transactions in 2025. Greg Ulrich said in Jul 2026 that generative AI lets Mastercard find 300% to 400% more fraud in high-risk bands without more false positives. | Security services | In production | |
| Threat intelligence and third-party risk Mastercard combines Recorded Future's threat intelligence, acquired in Dec 2024, with its fraud data to spot compromised cards and threat actors. In Q3 2026, Mastercard merged RiskRecon and Recorded Future capabilities into one third-party risk offering. | Security services | In production | |
| Agentic Trust Intelligence Agentic Trust Intelligence, launched shortly before the Oct 29, 2026 earnings call, helps banks and merchants evaluate transactions initiated by any AI agent. | Security services | In production | |
| Agent Pay Agent Pay lets AI agents that the buyer has verified and authorized pay with tokenized Mastercard credentials. Shopify's Shop Pay uses Agent Pay, and OpenAI agreed to use it under the Q3 2026 partnership. | Consumer Payments | In production | |
| Agent Pay for Machines Agent Pay for Machines handles high-frequency, low-value payments between agents and machines for services such as APIs, compute and data, with more than 30 launch participants. | Consumer Payments | Announced | |
| Mastercard Max benefits assistant Mastercard Max is a conversational assistant that surfaces and explains card benefits inside an issuer's own app, with Mastercard running the agent models and content. Mastercard Max launches first with C6 Bank in Brazil, and SNB Group, Nedbank, I&M Bank and Attijariwafa Bank are working to offer it (Oct 2026). | Consumer Payments | Announced | |
| Agent Suite for banks and merchants Agent Suite lets banks and merchants build, test and deploy agents for product discovery and conversational shopping with help from Mastercard's 4,000 advisers (Jan 2026). | Services | Announced | |
| Personalization services Mastercard's personalization platform and decision engine use AI to deliver product recommendations, offers and content to consumers across digital channels for its business customers (2025 Form 10-K). | Services | In production | |
| Large tabular foundation model Mastercard is testing a foundation model trained on billions of anonymized transactions as the base for hybrid cybersecurity systems and, later, loyalty, personalization and portfolio optimization tools. | Data and AI | Pilot | |
| Consultant research and analysis agents Mastercard's 4,000 consultants use in-house agents for deep research, text-to-SQL queries, Excel and PowerPoint, built in 2025. | Services | In production | |
| Product onboarding assistant Mastercard's onboarding assistant answers customers' questions about setting up Mastercard products from its own documentation, with expert review of its answers, as the first of a planned series of knowledge agents (Oct 2024). | Customer onboarding | In production | |
| AI coding assistants Ed McLaughlin said on Nov 9, 2026 that all of Mastercard's developers have AI assistance for software engineering and that some areas have seen throughput rise 50% with fewer defects. Mastercard developers began with GitHub Copilot and other assistants after a pilot with 2 small groups (Sep 2025). | Technology | In production | |
| Microsoft 365 Copilot and OpenAI tools for staff Mastercard rolled out Microsoft 365 Copilot to 16,000 employees by Sep 2025 and uses OpenAI's tools as an enterprise customer (Apr 2026). Mastercard's internal agents also help train new managers. | Firmwide | In production | |
| Vulnerability hunting with frontier models A Mastercard security team routinely runs the company's most critical assets through Anthropic's Claude Mythos and OpenAI's GPT-5.5-Cyber, ranks findings by severity and uses the same technology to patch them. Mastercard is working to offer the approach to others (Jul 2026). | Cybersecurity | Pilot |
Adoption
Mastercard spread generative AI to staff in steps, starting with mandatory data policy training before access and a review of each employee group's workflows at the end of 2024. Microsoft 365 Copilot reached 16,000 employees and coding assistants reached developers by Sep 2025, and OpenAI's enterprise tools followed by Apr 2026. Greg Ulrich said in Oct 2026 that more than 80% of Mastercard's employees use AI regularly, without saying how often. Ulrich said AI Learning Hub sessions had drawn more than 13,500 attendances. Ed McLaughlin said on Nov 9, 2026 that all of Mastercard's developers have AI assistance. Mastercard has not published weekly or daily active user counts for any internal tool.
By the numbers
Rollout, first pilot to today
Feb 2024
Mastercard announced Decision Intelligence Pro, applying generative AI techniques to the fraud scoring service that scored 143 billion transactions a year.
Apr 2024
Mastercard created a Data and AI organization under Greg Ulrich as Chief AI and Data Officer, overseen by Chief Services Officer Craig Vosburg.
Oct 2024
Mastercard launched a knowledge agent program built with Databricks, starting with a product onboarding assistant.
Dec 2024
Mastercard completed its acquisition of Recorded Future to add AI-driven threat intelligence to its security services.
Feb 2025
Mastercard appointed Janet George as executive vice president of artificial intelligence to lead its AI Center of Excellence.
Apr 2025
Mastercard launched Agent Pay with Microsoft and IBM as early partners.
Sep 2025
George Maddaloni said Mastercard had rolled out Microsoft 365 Copilot to 16,000 employees, GitHub Copilot to developers, and in-house agents for consultants and manager training.
Jan 2026
Mastercard announced Agent Suite to sell customizable AI agents and advisory help to banks and merchants.
Mar 2026
Mastercard described its large tabular foundation model, trained on billions of anonymized transactions with Nvidia and Databricks.
Apr 2026
Michael Miebach said nearly all Mastercards worldwide were enabled for Agent Pay and that Mastercard uses OpenAI's tools as an enterprise customer.
Jul 2026
Greg Ulrich described Mastercard's agentic factory, with guardrails and automated drift tracking, and its agents for 4,000 consultants.
Oct 2026
Greg Ulrich said more than 80% of Mastercard's employees use AI regularly.
Nov 2026
Ed McLaughlin said all of Mastercard's developers have AI assistance for software engineering.
Economics & controls
Spend and ROI
Technology and AI spending
Mastercard does not publish a technology budget or an AI budget. Mastercard's adjusted operating expenses rose 12% in Q3 2026, driven by spending on strategic priorities including product innovation and technology, and Ling Hai expected full-year 2026 operating expense growth in the low double digits. Futuriom reported in Nov 2025 that Mastercard matches each dollar it spends on new AI tools or licenses with an equal investment in training and support.
Reported ROI
Mastercard has reported no revenue, cost, unit cost or headcount figure that it credits to AI. Greg Ulrich said on Jul 14, 2026 that services make up about 40% of the company and that about a third of them are predicated on AI and growing faster than the rest. Ulrich told Forbes in Oct 2026 that Mastercard judges AI by workforce adoption, productivity, customer outcomes and market leadership, but gave no values. Mastercard's fraud figures, such as Decision Intelligence Pro lifting fraud detection by an average of 20% in initial modeling (Feb 2024), describe gains for the banks that buy the service.
Headcount
Mastercard employed about 39,800 people at Dec 31, 2025, plus about 6,000 contingent workers. Mastercard said on Jan 29, 2026 that a strategic review would lead to actions affecting about 4% of its full-time employees. Mastercard recorded a $202 million restructuring charge in Q1 2026 and plans to reinvest the savings in long-term growth. The company did not attribute the cut to AI.
Governance and risk
Board oversight
Mastercard's board oversees the company's use of AI and data, according to the 2026 proxy statement. The full board receives regular AI strategy updates from Chief AI and Data Officer Greg Ulrich, including on potential uses of generative AI and on generative AI governance principles. The board's Risk Committee receives periodic briefings from the Chief Privacy, AI and Data Responsibility Officer, Caroline Louveaux.
AI governance process
Mastercard applies Privacy by Design, Data by Design and AI Governance processes to products built on its data and AI. Mastercard also tests for bias when it uses advanced analytics and has generative AI guidelines (2025 Form 10-K and 2026 proxy statement). Greg Ulrich described in Oct 2026 a company-wide responsible AI program with risk assessments, AI inventories, model monitoring and human oversight. The program requires responsible AI training for all data scientists.
Guardrails built into agents
Greg Ulrich said in Jul 2026 that Mastercard learned to build guardrails, security, observability and accountability into agents from the start, and that adding guardrails to an agent after it is built is doomed to fail. Mastercard's agentic factory carries those controls for every internal agent and automates model drift tracking.
Data boundary
Mastercard requires employees to complete training on its data policies before they get access to generative AI tools (Sep 2025). Mastercard removes personal data from transactions before training its foundation model (Mar 2026). George Maddaloni said in Sep 2025 that Mastercard is unlikely to expose customer-facing agents through the Model Context Protocol because of security concerns.
Regulation and disclosed risks
Mastercard's 2025 Form 10-K says that as a user and deployer of AI, Mastercard is subject to laws including the EU AI Act. The 10-K says the regulation of Mastercard's AI use is still developing, especially in agentic commerce, and that criminals' use of AI has made fraud and cyberattacks worse.
Recent signals
Nov 9, 2026
Ed McLaughlin told Citi's FinTech Conference that 100% of Mastercard's developers have AI assistance for software engineering and that some areas have seen 50% higher throughput with fewer defects.
Oct 29, 2026
Michael Miebach said on the Q3 2026 call that Mastercard signed a broad strategic partnership with OpenAI covering Agent Pay, payments, joint services growth and AI transformation. Miebach said agentic transactions are growing quickly from a small base.
Oct 21, 2026
Greg Ulrich told Forbes that more than 80% of Mastercard's employees use AI regularly and that AI Learning Hub sessions had drawn more than 13,500 attendances.
Oct 21, 2026
Mastercard launched Mastercard Max, an AI benefits assistant for card issuers, starting with C6 Bank in Brazil.
Jul 30, 2026
Mastercard's Q2 2026 call highlighted Agent Pay for Machines and Merchant Trust Services, a suite of AI tools for spotting fraudulent merchants.
Jul 14, 2026
Greg Ulrich told VB Transform that Mastercard built an agentic factory for internal agents and joined the early wave of Anthropic's Project Glasswing, and that about a third of its services are predicated on AI.
Jun 10, 2026
Mastercard launched Agent Pay for Machines for agent-to-agent and machine payments, with more than 30 launch participants including Adyen, Coinbase and Stripe.
Jun 2, 2026
Mastercard named Ling Hai CFO and Linda Kirkpatrick Chief Services Officer from Aug 3, 2026, and put Jorn Lambert in charge of Consumer Payments, including agentic payments.
Apr 30, 2026
Michael Miebach said Mastercard is deepening its OpenAI partnership and using OpenAI's tools as an enterprise customer, and that nearly all Mastercards worldwide are enabled for Agent Pay.
Apr 27, 2026
Mastercard's 2026 proxy statement described board oversight of AI and said George Maddaloni, CTO of Operations, would leave on May 29, 2026.
Mar 17, 2026
Mastercard described a large tabular foundation model trained on billions of anonymized transactions with Nvidia and Databricks, shown at Nvidia GTC 2026.
Feb 11, 2026
Mastercard's 2025 Form 10-K reported about 39,800 employees and said Agent Pay reached all US cardholders in 2025, with a global launch set for early 2026.
Jan 29, 2026
Mastercard said a strategic review would lead to actions affecting about 4% of its full-time employees and a restructuring charge of about $200 million, without attributing them to AI.
Jan 27, 2026
Mastercard announced Agent Suite, customizable AI agents for banks and merchants backed by 4,000 advisers.
Quick answers
- How does Mastercard deploy AI?
- Mastercard builds AI into the services it sells: its models score transactions for fraud, and Agent Pay lets AI agents pay on its network. Inside the company, Mastercard runs AI through an agent platform that Chief AI and Data Officer Greg Ulrich calls an agentic factory. Ulrich said in Oct 2026 that more than 80% of Mastercard's employees use AI regularly. Mastercard has published no revenue, cost or headcount figure that it credits to AI.
- What does Mastercard use AI for?
- Mastercard's AI use cases fall into fraud and security services for banks and merchants, agentic payments on its network, personalization and consulting services, and internal tools for staff, consultants and developers. Mastercard has published no return it credits to AI for any of them. Mastercard's fraud figures measure detection gains for client banks, and Greg Ulrich's statement that about a third of its services are predicated on AI comes with no revenue figure. Mastercard reports no firm-level AI return.
- Who leads AI at Mastercard?
- AI at Mastercard is led by Greg Ulrich (Chief AI and Data Officer); Janet George (Executive Vice President, Artificial Intelligence); Caroline Louveaux (Chief Privacy, AI and Data Responsibility Officer). Ownership of AI on the org chart: Chief AI and Data Officer Greg Ulrich, on the Management Committee.
- What AI platform does Mastercard use?
- Mastercard builds its own AI, including fraud models such as Decision Intelligence and a large tabular foundation model trained on billions of transactions (Mar 2026). Mastercard's internal agentic factory has compliance, observability and guardrails built in (Jul 2026).
- Which AI models does Mastercard use?
- Mastercard trains its own transaction models with Nvidia and Databricks, gives staff Microsoft 365 Copilot, GitHub Copilot and OpenAI tools, and has tested Anthropic's Claude Mythos and OpenAI's GPT-5.5-Cyber for security work.
- How many Mastercard employees use AI?
- Greg Ulrich said in Oct 2026 that more than 80% of Mastercard's employees use AI regularly. Microsoft 365 Copilot had reached 16,000 Mastercard employees by Sep 2025.
- What return has Mastercard reported from AI?
- Mastercard has reported no business return it credits to AI in the last 18 months, for any use case or for the company as a whole. Greg Ulrich said in Jul 2026 that about a third of Mastercard's services are predicated on AI and growing faster than the rest, but Mastercard publishes no revenue, cost or headcount figure for them. Mastercard's fraud model gains accrue to the banks that use them. On rollout, Ulrich, who has owned AI as Chief AI and Data Officer since Apr 2024 and sits on the Management Committee, told Forbes on Oct 21, 2026 that more than 80% of Mastercard's employees use AI regularly. Ed McLaughlin said on Nov 9, 2026 that all of Mastercard's developers have AI assistance, and that rollout places Mastercard at Piloting. Mastercard falls short of Scaling because it reports no AI-credited return in any function and no firm-level AI figure.
Open questions
- How much revenue do Mastercard's AI-predicated services bring in, given that Greg Ulrich puts them at about a third of services but Mastercard publishes no figure?
- What have Mastercard's internal AI tools returned in cost, capacity or headcount, given that Greg Ulrich says Mastercard measures adoption and productivity but has published no values?
- How often do the more than 80% of Mastercard employees who use AI regularly use it, and which tools do they use?
- Which public cloud providers run Mastercard's AI models, including the large tabular foundation model, and how much runs in Mastercard's own data centers?
- Who leads the internal AI tool rollout after George Maddaloni's departure on May 29, 2026, and does Greg Ulrich's organization now sit under Chief Services Officer Linda Kirkpatrick?
- Does Janet George still lead the AI Center of Excellence, and how many data scientists and AI engineers does Mastercard employ?
- What does the Q3 2026 OpenAI partnership's AI transformation strand cover inside Mastercard?
- Did any of the roles cut in Mastercard's 2026 restructuring go because of AI?
- What did Mastercard find, in number and severity of vulnerabilities, when it ran its critical assets through Claude Mythos and GPT-5.5-Cyber?
- When will the large tabular foundation model move from testing into Mastercard's production fraud and cybersecurity services?
Sources
Mastercard
- Greg Ulrich, Chief AI and Data Officer, biography
- Ed McLaughlin, President and CTO, biography
- Michael Miebach, CEO, biography
- Jorn Lambert, Chief Product Officer, biography
- Linda Kirkpatrick, Chief Services Officer, biography
- Craig Vosburg, Vice Chair, biography
- Mastercard appoints Janet George as EVP of artificial intelligence (Feb 2025)
- Mastercard's new generative AI model (Mar 17, 2026)
- Mastercard launches Agent Pay for Machines (Jun 10, 2026)
- Mastercard launches Agent Suite (Jan 27, 2026)
- Mastercard Max makes benefits easy (Oct 21, 2026)
- Mastercard unveils Agent Pay (Apr 29, 2025)
- Mastercard press release on generative AI in Decision Intelligence Pro (Feb 2024)
- Mastercard launches new gen AI digital assistant capabilities (Oct 2024)
Filings and earnings calls
- Mastercard 2025 Form 10-K (filed Feb 11, 2026)
- Mastercard Form 10-Q for Q3 2026 (filed Oct 29, 2026)
- Mastercard 2026 proxy statement (Apr 27, 2026)
- Mastercard leadership updates, 8-K exhibit (Jun 2, 2026)
- Q3 2026 earnings call transcript, The Singju Post (Oct 29, 2026)
- Q2 2026 earnings call transcript, The Motley Fool via Webull (Jul 30, 2026)
- Q1 2026 earnings call transcript, Bullfincher (Apr 30, 2026)
- Q4 2025 earnings call transcript, Bullfincher (Jan 29, 2026)
- Ed McLaughlin at Citi's 15th Annual FinTech Conference, Investing.com transcript (Nov 9, 2026)
News
- Forbes: Mastercard is delivering business and customer value from data and AI (Oct 21, 2026)
- Yahoo Finance: syndicated copy of the Forbes interview with Greg Ulrich (Oct 21, 2026)
- VentureBeat: Greg Ulrich at VB Transform 2026 (Jul 30, 2026)
- Runtime: How Mastercard encouraged AI adoption with training and data (Sep 23, 2025)
- The Paypers: Mastercard partners with OpenAI on Agent Pay and AI transformation (Nov 2026)
- American Banker: Mastercard realigns business structure (Apr 2024)
- CDO Magazine: Mastercard names Greg Ulrich Chief AI and Data Officer (Apr 2024)
- Built In: What it's like to build generative AI models at Mastercard (Apr 2026)
- FX News Group: Mastercard closes acquisition of Recorded Future (Dec 2024)
- Futuriom: Enterprise AI profile, Mastercard (Nov 2025)
- Tech Brew: Mastercard's chief AI officer, Greg Ulrich (Aug 16, 2024)
- Forbes Belgium: Caroline Louveaux on AI at Mastercard (May 11, 2026)
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
Mastercard profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 9, 2026
