Key Takeaways
- Over $200 million was stolen via deepfake scams in Q1 2025, showing the scale of the threat.
- Scammers target celebrities, politicians, and everyday people alike—no one is off-limits.
- Detection tools and laws are lagging behind, leaving banks, tech firms, and users exposed.
In just the first quarter of 2025, scammers stole more than $200 million through deepfakes.
Of the 163 reported cases, 41% targeted celebrities and politicians, while 34% went after everyday people. And that’s only what’s been documented.
What used to be a problem for the famous is now a problem for everyone.
If you’ve got a social media presence, you could be a target.
Why “Good Enough” Isn’t Good Enough
Skeptics often argue that deepfakes aren’t a big deal.
People can spot obvious fakes, detection tools keep improving, and heavy regulation risks stifling AI innovation. But this optimism misses the point.
Researchers recently ran tests on the four biggest Language Models — ChatGPT, Claude, Gemini, and Grok.
They wanted to see how well these tools could spot deepfakes mixed in with real photos.
They found the models failed in predictable ways—getting distracted by visual style and tripped up by misleading patterns.
Even when they gave explanations, the results weren’t reliable enough for dependable deepfake detection.
That shows the “good enough” argument doesn’t hold, especially when criminals are siphoning millions through ever more sophisticated scams.
Every Post Is Ammunition
Every social media post is ammunition for scammers.
It takes only a handful of Facebook photos to generate a convincing fake video, and as little as 10–15 seconds of audio to perfectly clone a voice.
Criminals are stockpiling these scraps of online identity into massive databases of synthetic personas, fueling fraud operations that traditional security measures can’t contain.
Tech Isn’t Keeping Up
Banks, social platforms, and tech companies are only left to react.
Their detection tools are trained on yesterday’s deepfakes and are already failing against tomorrow’s threats.
A March 2025 study from Australia found that the best deepfake detectors identified fakes 86% of the time in lab settings—but dropped to just 69% in real-world tests.
The real issue isn’t the tech itself—it’s the mindset.
Too many companies treat detection as a one-and-done project. Meanwhile, scammers are already three steps ahead with newer tools.
What AI Defense Should Look Like
The military figured this out decades ago: assume your adversary is always getting smarter.
Update constantly. Plan for attacks that don’t exist yet. By contrast, tech companies still follow the old “build once, patch later” model, which fails in a rapidly evolving threat landscape.
True deepfake defense means layered safeguards. If a voice clone calls your bank to request a $50,000 transfer, that alone should never be enough.
Require text verification, additional questions, multiple checkpoints.
If a “CEO” demands an urgent wire transfer, pick up the phone and confirm through another channel. We’ve learned not to give Social Security numbers to random callers—now we must learn not to trust voices and faces at first glance.
The Limits of the AI Law
Denmark recently passed legislation banning the creation and distribution of AI-generated deepfakes without consent, protecting citizens from harassment and reputational damage.
This is a commendable first step—but one that falls short against professional scammers.
Unlike amateurs who use consumer apps, criminal groups operate on encrypted networks, host their own AI models, and launder money through offshore accounts.
Static laws built for trackable platforms can’t reach them. It’s like trying to fight organized cybercrime with old-fashioned copyright claims—it simply doesn’t work.
The Road Ahead
The $200 million lost in Q1 is just the beginning.
Even Sam Altman, CEO of OpenAI, has warned of an “impending fraud crisis,” admitting that AI has already broken most traditional authentication systems.
Unless tech companies, lawmakers, and everyday people act quickly, legal remedies that assume identifiable bad actors will become just as obsolete as the authentication systems criminals are already bypassing.
Technical solutions need to evolve in step with legal frameworks and public education. The question isn’t whether deepfake crimes will escalate—they will.
The real question is whether we’ll build detection systems that adapt as fast as the threats, or keep producing yesterday’s solutions for tomorrow’s problems.
Recommended Secure Partners
The views, thoughts, and opinions expressed in the article belong solely to the author, and not necessarily to CCN, its management, employees, or affiliates. This content is for informational purposes only and should not be considered professional advice.
About the Author
Ken Jon Miyachi is the co-founder of BitMind, a company at the forefront of developing pioneering deepfake detection technology and decentralized AI applications.
Prior to founding BitMind, Ken served as a software engineer and technical lead at leading organisations such as NEAR Foundation, Amazon, and Polymer Labs, where he honed his expertise in scalable technology solutions.
He has written several academic research publications on blockchain from his work at the San Diego Supercomputer Center.
