A woman in her old age in the UK got an unexpected call from her grandson. He was panicked and stranded, urgently required bail money. His voice was loud and clear. She wired the funds straight away.
That wasn’t her grandson’s call.
She heard a deepfake, an artificial intelligence generated voice that was created from a collection of public audio files she pulled off social media. When the family finally knew what had happened between the time had passed and the money was gone.
This isn’t a cautionary story set in some far-flung cyberpunk future. In 2026, deepfakes make up 11% of all fraud in the world, and the technology behind them is free and does not need technical skills and can be used anonymously. It’s that kind of combination that makes this threat so real that it’s not like any threat that we’ve seen before.
What Deepfake Technology Actually Does
Deepfakes are often mistaken for a manipulated video of a celebrity or a political disinformation. This approach is definitely a risky one.
Deepfake video generation has progressed from noticeable fakes to real-time interactive avatars. New models are not subject to the ‘flickering’ or ‘warping’ or ‘uncanny valley’ effects that previously required specialized techniques to be detected.
Today, deepfake technology operates across three attack surfaces most people are completely unprepared for.
Voice cloning requires as little as three seconds of audio to replicate someone’s voice convincingly. That audio can come from a voicemail, a social media video, a podcast appearance, or a work presentation. Once cloned, the voice can say anything in real time.
Video deepfakes can now run live during video calls. The Arup case demonstrated that deepfake video participants can fool experienced professionals in live calls. The engineering firm lost $25.6 million in a single incident where an employee was convinced they were speaking to their CFO.
AI-generated phishing is the quieter threat. LLM-powered phishing uses large language models to generate hyper-personalised emails that reference specific organisational details, recent transactions, and individual communication styles, attacks that lack the telltale signs legacy email filters were trained to catch.\
The Signals Most People Miss
The bad news is that the tell-tale signs of a deep fake are being systematically removed. But new ones have sprung up, and the knowledge of them changes everything.
Unusual eye blinking or eye motion. Even when it comes to sophisticated video deepfakes, they still struggle with producing natural eye behaviour. Blink rate should be regular, eyes track slightly and should reflect light consistently.
Non breathing audio. Real speech has micro pauses, breath sounds and irregular cadence. Cloned voices may sound a bit too robotic, a bit too perfect, a bit too robotic and perfect. If something sounds like a terrific voicemail recording, instead of a real conversation, that is worth noting.
Requests that are not through your normal process. This is the best indicator of all and is based on nothing more than technology. The deepfake scams always create a sense of urgency and request you to do something that you wouldn’t normally do. A CFO who suddenly requires a wire transfer via WhatsApp.Suddenly. A grandchild who is unable to contact them on their usual phone number. A computer system that requires your signature for access. Legitimate contacts don’t pressure you to go around the process.
Inconsistencies in the foreground and background. When it comes to subjects, video deepfakes tend to perform well on them, but do less than optimal on hair edges, glasses, or anything near the edge of the human and environment. If the video call has soft edges or the face is slightly blurry, it is a significant red flag.
What Artificial Intelligence Is Now Doing to Protect You
The same artificial intelligence driving deepfake scams is being turned against them, and the results are significant.
Avast has built scam and deepfake detection directly into its security suite. Avast One offers robust protection against cybersecurity threats including AI-generated scams, malware, viruses, deepfakes, and more, and helps identify suspicious texts and links using its built-in AI assistant. The avast malware protection layer runs continuously in the background, which means it catches threats before they reach you rather than after damage is done.
For anyone looking to understand exactly what avast antivirus covers across different devices, the Avast One plans page breaks down the full feature set clearly, including the scam protection tools that are most relevant to the deepfake threat.
Norton 360 has made deepfake detection one of its headline 2026 features. Norton 360 includes AI-powered scam protection that detects phishing in texts, emails, and deepfake videos: a 2026 addition that competitors have not matched. The norton identity theft protection layer goes further, monitoring dark web activity for personal data exposure and alerting you before your information is weaponised against you.
Norton’s deepfake detection works automatically on supported platforms, flagging manipulated video content in real time. The specifics of what each norton 360 plan includes and how the deepfake detection feature actually functions are laid out in detail on the Norton AI scam protection page, which is one of the more honest and thorough explanations of how this technology works in practice.
Malwarebytes takes a different approach that deserves more credit than it typically gets. Malwarebytes Plus offers modern cyber safety tools including its AI-powered Scam Guard and Digital Footprint Scanner, along with a very generous 60-day money-back guarantee.
The malwarebytes free vs premium question comes down to one practical distinction: the free version handles detection and removal after the fact, while the premium version blocks threats proactively before they can execute. For anyone weighing malwarebytes pricing against what they actually get, the Malwarebytes plans comparison page lays out the difference between tiers clearly, including the Scam Guard feature that is specifically relevant to AI fraud protection.
Five Habits That Make You a Harder Target
Technology helps, but habits remain your first line of defence.
Verify through a different channel. Urgent request from someone you know? Call them back on a number you already have saved. Never use the contact details provided in the suspicious message itself.
Create a family safe word. A short phrase agreed on in advance that anyone can use in a suspicious call to confirm they are real. Simple, free, and surprisingly effective.
Slow down when urgency is manufactured. The moment a request feels pressured, treat that pressure as the red flag. Deepfake scams run on speed. Take it away and the whole thing falls apart.
Keep your security software current. Deepfake detection relies on up-to-date threat databases. An outdated security suite is fighting last year’s attacks while this year’s are already running.
The Bottom Line
The number of deepfakes in the UK has increased by 94%, and the amount of fraud has remained the same in one year. Less attacks, more intense, targeted and automated.
If you know how deepfake technology works, are able to spot the signs and have the proper protection running in the background without you even realising it, you’re in a very different position than many people. Immunity cannot be achieved through immunization. However, you can make yourself a much bigger target.
Frequently Asked Questions
Deepfakes scams are AI-powered technology that imitates voices, faces and video in real time to trick people they know and trust. It is a technology that demands low-skills and can bring realistic looking results from mere seconds of source audio or video.
Watch for unnatural blinking, hazy outline of the face and hair, an unusual level of smoothness in the voice, and an emergency request to perform something that you don’t normally do. If in any doubt, hang up the phone and check via another route.
Yes. Avast antivirus also features AI-based scam detection, which detects deepfakes, suspicious links and malicious content in real time. Avast runs in the background, providing protection for devices at all times.
Norton 360 comes with a dedicated deepfake detection feature, which automatically scans video content on supported platforms. The Norton identity theft protection also keeps an eye out for exposures of personal data on the dark web, providing a double defense against AI fraud.
Artificial intelligence allows scammers to clone voices from short audio clips, generate personalised phishing emails, create real-time video deepfakes, and run automated fraud operations at scale with minimal human involvement.
