Can AI Beat AI? Inside the New Fraud Arms Race

Generative AI has rewritten the rules of digital trust. It can write like us, speak like us and increasingly, pretend to be us.

That same capability now fuels an unprecedented surge in AI-driven fraud. In this ongoing bot-versus-bot battle, verification systems must do more than simply match identities in the traditional sense because AI has changed the game. Organizations now face the challenge of embracing cutting-edge technology to fight fire with fire and create an ecosystem of trust that ensures users are who they say they are in every interaction.

At Money20/20 USA, Trulioo Chief Technology Officer Hal Lonas moderated a powerhouse panel featuring compliance and technology leaders from Kraken, Worldpay and Meta to unpack the new AI battlefield: where machine-speed fraud meets machine-speed defense.

Their conversation revealed a new truth about the future of trust: in an era where fraud learns in real time, defense has to think faster, act smarter and collaborate more deeply than ever before.

1. When Fraud Learns Fast, Defense Must Think Faster

From onboarding to transaction monitoring, AI is now embedded in every corner of digital risk. But the same generative models that enable faster detection are also being weaponized to automate and personalize attacks.

This creates a cycle of AI versus AI – as soon as defenders deploy new models, attackers re-train theirs to adapt. The result? Static checks and siloed systems can’t keep up.

Ross Green, Global Head of KYC at Meta, highlighted the need to shift from reactive to proactive defense. His teams actively test their own systems using AI-powered simulations that mimic real-world attacks, helping them find and fix weaknesses before bad actors can exploit them.

“You can’t have static transaction-monitoring rules anymore,” Green said. “The defense has to learn as fast as the offense.”

The takeaway: AI defense isn’t a product – it’s a process. It depends on live feedback loops, cross-functional signal sharing and a culture of constantly testing and improving your own models before bad actors do.

2. Agentic Commerce Demands a New Layer of Trust: Know Your Agent

The next evolution of digital transactions won’t always involve humans directly. Intelligent agents are already researching products, placing orders and managing accounts.

Worldpay Chief Product Officer Cindy Turner cited a useful framework for this new reality:

Authenticate the agent: Is the AI or automation interacting with your system recognized and trusted?
Authenticate the user behind it: Has the person or organization controlling that agent been verified?
Authenticate intent: Do the agent’s actions truly align with what the user authorized?

As highlighted in the recent Worldpay–Trulioo announcement, the two companies are collaborating to advance the Know Your Agent (KYA) framework, combining Worldpay’s payments expertise with Trulioo identity technology to create the trust infrastructure that will keep AI-driven transactions secure, transparent and scalable.

“Payments aren’t the problem,” Turner noted. “Proving intent is the new frontier of trust.”

The solution lies in pairing strong authentication protocols with evolving standards that can verify the identity, origin and intent of AI agents, ensuring that as commerce becomes more autonomous, trust keeps pace.

3. Collaboration and Continuous Verification Will Decide Who Wins

Fraudsters operate across borders, verticals and data silos, often exploiting the gaps between organizations’ defenses. Amidst rising risk levels, no company can fight AI fraud alone.

Existing frameworks allow financial institutions to share intelligence without violating privacy laws. But responsible collaboration, within industries and across them, will define the next generation of fraud defense.

David Zacks, deputy chief compliance officer at Kraken noted: “Programs today are built to meet regulations but they’re not always built to stop bad actors.”

Internally, that collaboration starts with unifying fraud, compliance and security signals. Externally, it means secure data-sharing protocols, common taxonomies and open standards for real-time threat intelligence.

When combined with continuous verification, collaboration becomes a competitive advantage. The organizations that share intelligence responsibly will move faster, detect earlier and build stronger trust ecosystems than those that stand alone.

Turning AI Into a Force for Trust

The arms race between AI-driven fraud and AI-powered defense is only accelerating. But rather than a battle of machines, it’s a test of how quickly organizations can adapt their systems, standards and ways of thinking. The winners won’t be those with the biggest models, but those with the smartest trust architecture: continuous, collaborative and built to evolve at machine speed.

That evolution starts with identity. Know Your Agent (KYA) extends the principles of digital trust into the world of autonomous systems, creating a foundation where human and machine interactions can be authenticated, authorized, and accountable.

Because the question isn’t just whether AI can beat AI. It’s whether trust can move as fast as innovation – and with the right architecture, it already is.