How Gen and Intel are Redefining Deepfake Detection AI

Cyber safety leader Gen – the parent company of Norton – has unveiled an early look at its upcoming on-device deepfake detection technology at CES, marking a major step forward in tackling AI-driven scams.

Developed in collaboration with Intel, the new feature identifies both audio and visual manipulation in real time, directly on user devices, delivering faster and more private protection without relying on cloud processing.

The reveal coincides with new insights from Gen that challenge prevailing ideas about how and where deepfake scams are taking hold.

Instead of spreading through brief viral clips, these scams are more likely to surface during longer viewing sessions, where they merge seamlessly with regular content and gradually build credibility and influence.

“The presence of a deepfake alone is not the risk,” Vincent Pilette, CEO of Gen, says. "Risk emerges when deepfake capabilities are paired with intent: urgent financial requests, promises of guaranteed returns, pressure to act quickly, or instructions to move conversations or payments off platform."

Gen’s data shows that YouTube generates the highest proportion of intercepted deepfake-driven scam activity, followed by Facebook and X – platforms that are central to long-form, algorithm-driven viewing on TVs and PCs.

The United States records the greatest number of intercepted scam videos, with Germany, the United Kingdom, Canada and Australia rounding out the top five.

Are deepfake scams hiding in plain sight?

Gen reports that the majority of deepfake scam videos are identified during playback, rather than through downloads, shared links or attachments.

“They hide in plain sight, embedded in ordinary video consumption,” Gen says.

This contrasts sharply with traditional phishing techniques, where malicious content typically appears through more obvious channels such as suspicious emails or text messages.

Gen also notes that audio-led deception has become a dominant element in the threat landscape. 

Many scams depend on cloned or synthetic voices, combined with subtly altered visuals from legitimate video footage – much like dubbed foreign-language videos.

This, Gen explains, makes detection especially difficult, as the visuals may seem entirely genuine while the audio delivers the fraudulent narrative.

“The tools AI creators rely on for voice cloning, synthetic narration, automated editing and composite visuals, are now standard capabilities across many popular platforms and production workflows,” Vincent adds.

"Adobe’s latest global survey found that 86% of creators use generative AI somewhere in their process. Analysts expect a growing share of outbound marketing messages to be synthetically generated, enabling persuasion at scale, delivered faster and with fewer human bottlenecks.”

Unsurprisingly, financial enticements continue to dominate, with investment tips, trading schemes, cryptocurrency offers and giveaway promotions making up the majority of scam themes.

However, Gen stresses that deepfakes themselves are not inherently harmful – the real danger lies in how the technology is manipulated and weaponised.

“Our Deepfake Protection is built to stop these financial scams that cause the most harm – but it also protects against other scam types, grouping them into a broader scam category,” Vincent shares.

“Most blocked videos fell into this group, with financial and cryptocurrency following closely behind.”

The scale of the deepfake problem

Gen’s deepfake protection technology has advanced quickly since the company first introduced on-device detection of AI-generated audio in 2025 through its Norton product line.

Initially launched on AI PCs powered by Intel and Qualcomm, the capability has since expanded to standard PCs and, at CES, achieved a new milestone.

Leveraging Intel’s forthcoming Panther Lake processor and Gen’s image analysis tool, the system can detect manipulated videos featuring public figures directly on the device – what Vincent describes as “a new benchmark for the industry.”

“Over time, our technology is expected to expand to detect more than just celebrities and famous people – it will also help to protect against more devastating scams such as family member impersonation,” he says.

This development comes amid an explosion of Gen AI in content creation – giving cybercriminals access to the very same tools once reserved for legitimate use.

Vincent says: “Criminals do not need a million views. They need the right viewer at the right moment.”

By focusing on long-form video analysis and real-time, in-playback detection, Gen aims to spot scams as they happen rather than after the damage is done.

The company’s strategy zeroes in on malicious intent rather than AI creativity, helping rebuild trust as deepfakes become more widespread and embedded in mainstream media.