January 22, 2026
Today's Briefing: January 22nd 2026
NewsAgents Take
Today’s material reinforces that agentic commerce is advancing unevenly and pragmatically. Adoption is being shaped less by model capability and more by trust, fraud exposure, and existing payment infrastructure. Data from the Global Payments 2026 report shows strong awareness but limited readiness among both merchants and consumers, while discussions from the MPE webinar highlight how AI is already amplifying fraud risk in ways that complicate automation. The picture that emerges is not one of stalled progress, but of a market that is rightly cautious as the cost of mistakes becomes clearer.
MPE webinar on agentic commerce and fraud risk
The MPE webinar focused on how AI and early agentic behaviour are intersecting with fraud and payments operations today. A recurring theme was that most merchants are experimenting with agentic concepts at the edges of the customer journey rather than delegating full transaction execution.
Speakers highlighted that the hardest problem is not intent discovery, but safe completion. As automation compresses decision timelines, traditional human review windows shrink. Fraud risk is shifting away from obvious misuse toward more subtle failures in intent interpretation, handoff logic, and authentication flows.
For payments and risk teams, the takeaway was clear. Agentic systems will only scale where controls, auditability, and rollback mechanisms are already strong.
https://www.youtube.com/watch?v=BtFlizwEybc
AI is expanding the payments fraud surface
Payments Dive reports that AI-enabled fraud is accelerating across payments channels. Fraud teams are seeing higher volumes of AI-assisted scams, including synthetic identity creation, automated social engineering, and faster enumeration attacks.
The article notes that issuers and merchants are increasingly investing in behavioural analytics and real-time anomaly detection, as static rules struggle to keep pace with adaptive threats. This matters directly for agentic commerce. As agents act faster and with less friction, the margin for error narrows and the cost of false positives and false negatives increases.
https://www.paymentsdive.com/news/payments-fraud-risks-burgeon-with-ai/810078/
What the Global Payments 2026 report says about agentic readiness
The Global Payments Commerce and Payment Trends Report provides a useful data-backed view of where agentic commerce stands today.
On the merchant side, 87 percent of businesses say they are at least somewhat familiar with the concept of agentic commerce, but only 15 percent describe themselves as very familiar. This gap between awareness and depth of understanding shows up in deployment readiness.
Consumer data is even more cautious. Only 11 percent of consumers report ever allowing an AI system to complete a purchase on their behalf. Just 4 percent say they would be interested in doing so today. Trust, security, and dispute handling are the dominant concerns.
On the risk side, 42 percent of businesses say they would be concerned if AI agents started purchasing from them now, with fraud and chargeback exposure cited most often.
The report also references external estimates suggesting agent-driven commerce could increase global ecommerce conversion by 1.5 to 2.5 percent, equivalent to roughly $240 billion in incremental revenue. However, that upside is explicitly tied to improvements in identity, authentication, and payment controls.
In short, the opportunity is real, but conditional.
https://www.globalpayments.com/commerce-payment-trends
Mastercard expands engagement with agentic startups
Mastercard’s decision to expand its startup engagement program to include agentic commerce is another signal that incumbents are preparing for agent-driven transactions, but on their own terms.
The focus of the program is less about agent interfaces and more about enabling infrastructure. Identity, authentication, tokenisation, and payment assurance remain the priority. This aligns with a broader pattern where networks and processors are positioning themselves as control layers rather than experience layers.
From agentic chaos to data discipline
MIT Technology Review frames the current phase of agentic AI as one of chaos driven by rapid capability gains without matching structure. The article argues that data quality, governance, and interoperability will ultimately determine whether agentic systems become reliable tools or persistent sources of risk.
For commerce and payments, this resonates strongly. Transactions are binary, disputes are costly, and trust is slow to rebuild. Agentic systems will only be welcomed where data discipline, consent, and accountability are built in from the start.
