Generative AI has changed the economics of deception. What previously required professional tools and hours of editing can now be done in just a few clicks. A realistic fake face, cloned voice or even a full video identity can be generated in minutes and used to pass verification systems that once seemed foolproof.
Over the past year, I have seen deepfake-based fraud accelerate at a pace that most organizations are unprepared for. Deepfake content on digital platforms increased by 550% between 2019 and 2024 and is now considered one of the top global risks in today’s digital ecosystem. This isn’t just a technological shift: it’s a structural challenge in how we verify identity, authenticate intent, and maintain trust in digital finance.
Adoption goes beyond security
Crypto adoption in the United States continues to grow, fueled by increasing regulatory clarity, strong market performance, and increased institutional participation. The approval of spot Bitcoin ETFs and clearer compliance frameworks have helped legitimize digital assets for retail and professional investors. As a result, more Americans are viewing crypto as a dominant investment class – but the pace of its adoption still outpaces the public’s understanding of risk and security.
Many users still rely on outdated verification methods, designed for an era when fraud meant a stolen password, not a synthetic person. As AI generation tools become faster and cheaper, the barrier to entry for fraud has fallen to almost zero, while many defenses have not evolved at the same speed.
Deepfakes are used in everything from fake influencer livestreams that trick users into sending tokens to scammers to AI-generated video IDs that bypass verification checks. We’re seeing an increase in multimodal attacks, in which fraudsters combine doctored videos, synthetic voices, and fabricated documents to create entire fake identities that withstand scrutiny.
As journalist and podcaster Dwarkesh Patel noted in his book “The Scaling Era: An Oral History of AI, 2019-2025,” this is now the era of large-scale fraud. The challenge is not just sophistication, but also scale. While anyone can create a realistic fake with consumer software, the old model of “spotting the fake” no longer works.
Why current defenses fail
Most verification and authentication systems still rely on superficial cues: eye blinks, head movements, and lighting patterns. But modern generative models reproduce these micro-expressions with near-perfect fidelity – and verification attempts can now be automated with agents, making attacks faster, smarter and harder to detect.
In other words, visual realism can no longer be the gold standard for truth. The next phase of protection must go beyond what is visible and focus on behavioral and contextual cues that cannot be imitated. Device models, typing rhythms, and micro-latency of responses are becoming the new fingerprints of authenticity. Eventually, this will extend to some form of physical authorization – from digital IDs to implanted IDs, to biometric methods like iris or palm recognition.
There will be challenges, especially as we become more comfortable allowing autonomous systems to act on our behalf. Can these new signals be imitated? Technically, yes – and that’s what makes it a continuing arms race. As defenders develop new levels of behavioral security, attackers will inevitably learn to replicate them, forcing constant evolution on both sides.
As AI researchers, we must assume that what we see and hear can be fabricated. Our task is to find the traces that manufacturing cannot hide.
The next evolution: trust infrastructure
Next year will mark a turning point for regulation as trust in the crypto sector remains fragile. While the GENIUS Act has become law and other frameworks like the CLARITY Act are still under discussion, the real work is filling the gaps that regulation has not yet filled – from cross-border enforcement to defining what meaningful consumer protection looks like in decentralized systems. Policymakers are beginning to establish rules on digital assets that prioritize accountability and security, and as new frameworks take shape, the industry is gradually moving toward a more transparent and resilient ecosystem.
But regulation alone will not solve the trust deficit. Crypto platforms must adopt proactive, multi-layered verification architectures that don’t stop at onboarding, but continuously validate the identity, intent, and integrity of transactions throughout the user journey.
Trust will no longer depend on what seems real but on what can be proven. This is a fundamental change that redefines the infrastructure of finance.
A shared responsibility
Trust cannot be modernized; this must be integrated. Since most fraud occurs after onboarding, the next phase depends on moving beyond static identity controls to continuous, multi-layered prevention. Connecting behavioral signals, cross-platform intelligence and real-time anomaly detection will be key to restoring user trust.
The future of crypto will not be defined by how many people use it, but by how many people feel safe doing so. Growth now depends on trust, responsibility and protection in a digital economy where the line between real and synthetic continues to blur.
At some point, our digital and physical identities will need even greater convergence to protect us from imitation.
