AI agents power UPI payments—but fraud battle intensifies

The National Payments Corporation of India (NPCI) is spearheading a transformative leap in digital payments with the launch of its Unified Agent Protocol (UAP) at GFF 2026, designed to enable AI-driven, autonomous UPI transactions that incorporate intent-aware authentication. This protocol not only promises to streamline payments by understanding user intent but also integrates enterprise workflow automation, signaling a strategic move toward more intelligent, seamless financial interactions across India's vast digital payments ecosystem. Complementing the UAP, NPCI has introduced two additional AI-powered platforms aimed at automating complex enterprise workflows, underscoring its commitment to embedding agentic AI capabilities that enhance operational efficiency and user experience within the UPI framework.

The lightning-fast nature of UPI transactions, clearing in under a second, drastically compresses the fraud detection window from the traditional 15 minutes to as little as 30 seconds, leaving conventional reactive fraud detection methods woefully inadequate. This acceleration demands a paradigm shift from detecting fraud after the fact to predictive risk intelligence that continuously monitors behavioral signals across merchants, devices, customers, and networks to identify anomalies before fraudulent transactions occur.

As criminals harness generative AI to craft highly convincing phishing messages, synthetic identities, and deepfakes, the sophistication and scale of fraud have surged dramatically, with India witnessing a tenfold increase in reported digital payment fraud cases from 2.6 lakh in 2021 to 28 lakh in 2025, and losses soaring over 40 times to ₹22,931 crore. This evolution challenges traditional credential validation, pushing financial institutions to verify consistent behavioral patterns and deploy layered verification combining identity, device, behavioral, transactional, and contextual signals to distinguish legitimate AI-driven agentic transactions from malicious automation.

To combat the compressed fraud timeline and sophisticated attack vectors, banks and fintechs are increasingly relying on advanced AI/ML-powered tools like RBI Innovation Hub’s MuleHunter.AI and the Financial Fraud Risk Indicator (FRI), which have collectively helped prevent suspected fraud losses exceeding ₹5,043 crore by August 2026. These solutions integrate real-time risk signals directly into transaction routing and promote cross-institution data sharing, enabling a dynamic, multimodal defense that combines transaction intelligence with behavioral biometrics, device intelligence, and deepfake detection, while retaining human oversight to ensure accountability and regulatory compliance.

The rise of agentic UPI payments—where AI systems initiate or complete transactions autonomously—introduces new complexities for fraud detection models traditionally built around human behavior. Distinguishing between legitimate machine-driven actions and malicious automation requires evolving fraud prevention frameworks to incorporate continuous adaptive AI systems that analyze network, device, and behavioral data holistically within seconds, ensuring that the defense keeps pace with the accelerating AI-powered arms race between attackers and defenders.

Kurv CTO Mark Alsentzer envisions a transformative shift in SMB payments where AI moves beyond mere automation to proactively simplifying and personalizing the merchant experience. By eliminating outdated manual rituals like logging into dashboards, Kurv aims to build robust data architectures that enable AI-driven features delivering near-term value without compromising security. This proactive approach not only streamlines payments but also tailors financial services to the unique needs of small and medium-sized businesses, marking a significant evolution in how merchants interact with payment systems.

Behind the scenes, AI is revolutionizing core payment processes such as transaction routing, approval decisions, and reconciliation, which collectively enhance the seamlessness and reliability of payment experiences for merchants. Alsentzer highlights that these improvements, though invisible to end users, are critical in elevating the overall efficiency and satisfaction in SMB payment workflows, demonstrating how AI’s subtle orchestration can yield outsized benefits.

The emerging 'centaur' model of human-agent collaboration in financial services exemplifies how AI agents complement human decision-making by proactively managing portfolios, surfacing opportunities, and monitoring risks in real time. This partnership not only simplifies financial management but also personalizes it, as agents assist rather than replace humans, signaling a future where AI-driven payment experiences are more intuitive and responsive to individual needs.

Looking further ahead, AI agents may redefine value exchange by prioritizing compute resources as a form of abstracted value, potentially shifting payment paradigms beyond traditional currencies. This conceptual evolution suggests a future marketplace where agents trade compute power directly, hinting at transformative implications for how payments and marketplaces operate in an increasingly AI-driven economy.

Experian is spearheading a fundamental shift in identity verification for AI-driven commerce by moving beyond traditional KYC to a 'Know Your Agent' (KYA) framework, which focuses on authenticating the AI agents themselves rather than just the human users. This approach is bolstered by innovations like Experian Agent Trust and digital identity tokens, which establish a secure, interoperable trust layer that bridges consumers, merchants, and financial institutions, thereby addressing the unique challenges posed by autonomous agents in payments.

Interoperability and strategic partnerships are central to building a scalable and trusted agentic commerce ecosystem. Experian’s collaboration with protocols such as KYA and X402 for micropayments, alongside tech partners like Cloudflare and Skyfire, ensures flexibility and adherence to open standards. Meanwhile, alliances with established financial institutions like American Express, whose agentic Commerce Experience developer kit integrates intent intelligence to discern malicious behavior, provide end-to-end transaction assurance and amplify trust by standing behind chargebacks, creating a robust framework for risk mitigation in AI-driven payments.