The era when humans can identify a deepfake is over, and it’s time to stop pretending otherwise.

by Tim Leogrande, BSIT, MSCP, Ed.S.

🗓  JULY 9 2026 • 5 MIN 30 SEC READ

A deepfake is a highly realistic piece of digitally manipulated media created using artificial intelligence to make it appear that a real person did or said something they never actually did or said. By analyzing authentic video, audio recordings, and images, AI platforms can convincingly replicate an individual’s face, expressions, movements, and voice. These profiles can then be used to generate fake videos, ads, interviews, photographs, and audio recordings that feature a digital doppelgänger of that person.

While deepfake technology has legitimate uses in entertainment, filmmaking, accessibility, and education, it’s frequently being deployed for fraud, identity theft, political disinformation, harassment, and other forms of deception. This makes it more difficult than ever to distinguish authentic content from convincing fabrications.

In April, a Copyleaks investigation reported a surge of TikTok ads using AI-generated deepfakes of celebrities. The likenesses of Taylor Swift, Rihanna, and Kim Kardashian were appropriated from real interview, newsreel, and performance footage. In one example, a digitally generated Taylor Swift claims she found a feature called “TikTok Pay” and redirects viewers to third-party sites that harvest personal data. Swift is arguably the most deepfaked likeness right now, with McAfee ranking her as the most impersonated celebrity across all Internet scam campaigns.

<aside> 💡

For years, the common advice for spotting deepfakes was simple. Look closely and watch for unnatural facial expressions, mismatched lip movements, strange lighting, or robotic speech. But advances in generative AI have reached the point where many deepfakes appear authentic not only to casual observers, but also to trained professionals. The result: organizations that still depend on human judgment as their primary defense against identity fraud are increasingly vulnerable, and deepfake-enabled scams are proliferating.

</aside>

According to the Identity Fraud Index report from Shufti, 2026 is on track for an approximately fivefold increase in deepfake identity fraud compared to 2025. The report tracked global fraud attempt data and projected huge growth in document deepfakes and synthetic identity fraud. Additionally, the Deloitte Center for Financial Services predicts:

"Generative AI is expected to significantly raise the threat of fraud, which could cost banks and their customers as much as US$40 billion by 2027."

The growth of generative AI tools has made this type of fraud dramatically easier for threat actors to perpetrate, so it no longer takes hours of work and a high level of expertise to craft a convincing fake identity. It can now require as little as a single image or a simple AI text prompt. As these platforms become more sophisticated, the technical barrier to entry for would-be scammers continues to fall.

Shufti tracked four primary areas of deepfake fraud:

• Synthetic identity: Entirely synthetic faces of people who don’t exist, which totaled 42.3 percent of 2025 deepfake fraud. This fraud type is expected to grow by around 73 percent this year.
• Live video deepfakes: Manipulated live video streams, which accounted for 28.1 percent of 2025 deepfake fraud.
• Face swaps: A victim’s face mapped onto an attacker’s head in real time, which was 17.6 percent of 2025 deepfake fraud.
• Document deepfakes: AI-produced documents were responsible for 11.9 percent of 2025 deepfake fraud. During 2026, this category is expected to grow to 40 times the 2025 total.

Dr. David Maimon, head of fraud insights for SentiLink, reports:

"We're seeing deepfakes used to replace faces, generate entirely new ones, and insert them into fake documents and liveness videos. Looking ahead, there is growing evidence that agentic AI systems may engage in deceptive techniques — and even autonomous hacking behaviors — without explicit human prompts."

The most effective deepfake attacks combine different techniques rather than relying on a single method. Attackers may blend a presentation attack with a synthetic identity or use an injection attack to bypass the camera entirely, piping an AI-generated identity directly into a verification system. Layered attacks make fraud far more difficult to detect because each technique helps reinforce the credibility of the others.