Who wins the AI vs AI battle in the agentic era?

/ 3 min read
AI Hub

The AI-versus-AI battle will not be won by the smartest algorithm. It will be won by the organisations that can make trust work at machine speed.

AI-generated
Credits: AI-generated

Every major shift in commerce has reshaped the nature of fraud. What began as a human-versus-human contest is rapidly becoming an AI-versus-AI battle.

ADVERTISEMENT

For decades, fraud relied on deception by individuals. Stolen cards, forged signatures, fake identities and social engineering schemes were used to exploit trust in the system. Defences were equally human led, with manual reviews, verification checks and investigative teams working to identify suspicious activity.

As commerce became digital, both attacks and defences evolved. Fraudsters started using malware, credential theft, automated bots, and account takeover techniques to operate with greater speed and reach. In response, organisations deployed machine-driven tools, such as behavioural analytics, real-time monitoring and machine learning models to identify suspicious activity and stop fraud before it caused harm.

ADVERTISEMENT

Today, we are entering another inflection point. Generative AI is transforming both sides of the equation. Deepfakes can convincingly replicate faces and voices. Synthetic identities can pass as legitimate users. AI-powered scams can adapt, and scale faster than traditional fraud techniques ever could.

But the same technology is also strengthening our ability to respond. Advanced AI systems can analyse billions of signals, detect anomalies in real time, and predict risks before fraud occurs. The battle is no longer human versus machine. Increasingly, it is AI versus AI.

 

The next phase of the AI arms race

This shift becomes even more consequential as commerce enters the era of AI agents. According to Accenture, by 2030, more than 30% of online or digital commerce could run through AI-driven platforms and agents, representing trillions of dollars in transactions.

Recommended Stories

Consumers will increasingly rely on AI-powered assistants to search, compare, recommend, and even purchase products on their behalf. This represents far more than a technological upgrade. It fundamentally changes how confidence is established in digital commerce.

Traditionally, fraud controls focussed on the individual making a transaction. Is the customer genuine? Is the payment credential valid? Does the transaction fit expected behaviour? In an agentic world, those questions remain important, but they are no longer enough.

ADVERTISEMENT

As AI begins to act on behalf of consumers, the focus extends beyond verifying users to understanding whether the systems initiating actions are authentic, operating within approved parameters, and drawing on trusted information. The boundaries of digital trust are expanding from people to the technologies acting for them.

Rethinking security for intelligent commerce

The urgency of this challenge is already becoming evident. According to the 2026 Thales Data Threat report, 65% of organisations in India reported deepfake-related incidents. At the same time, the Indian Computer Emergency Response Team (CERT-In) handled nearly 2.94 million cyber incidents in 2025, underlining the growing sophistication and scale of threats facing businesses and consumers alike.

Most Powerful Women In Business 2026
View Full List >

In an agentic ecosystem, the risks become even more complex. A compromised shopping agent could be manipulated into directing users to fraudulent merchants. Corrupted data sources could influence purchasing decisions. Impersonation attacks could trick systems into accepting unauthorised instructions. The consequences extend well beyond payment fraud, potentially affecting every stage of the customer journey.

Defending at machine speed

The encouraging reality is that defenders are not standing still. Mastercard’s Trust at Machine Speed report finds that AI systems now perform 50% of entry-level cybersecurity tasks, up from just 9% three years ago. By combining AI with network intelligence, we can now analyse identity, device, behavioural and transaction signals simultaneously, enabling faster detection of scams, impersonation attempts and other emerging fraud patterns.

This capability matters because the economics of cybercrime are changing. AI allows attackers to launch highly targeted campaigns at unprecedented speed and scale. The window for detecting and responding to threats is shrinking, making machine-speed defence increasingly essential.

Technology will remain only one part of the equation. The most effective defences are likely to emerge where AI-powered detection is supported by strong governance, quality data and closer coordination across cybersecurity, fraud and risk functions. As attacks become faster and more adaptive, resilience increasingly depends on how well these capabilities work together.

ADVERTISEMENT

The organisations that will win

Every generation of commerce introduces new risks. The businesses that emerge strong are rarely those that eliminate risk altogether, but those that adapt to changing threats while maintaining confidence across the ecosystem.

The same principle will define the agentic era. AI-powered threats will continue to evolve. AI-powered defences will evolve alongside them. But technology alone will not determine the winner.

ADVERTISEMENT

As AI becomes more deeply embedded in commerce, competitive advantage is likely to come from combining innovation with security, transparency and accountability.

In the end, the AI-versus-AI battle will not be won by the smartest algorithm. It will be won by the organisations that can make trust work at machine speed.

ADVERTISEMENT

(The author is Senior Vice President, Advisors & Consulting Services, South Asia at Mastercard. Views are personal.)

Follow Fortune India
NEXT STORY