NetGuardian Pro

Split-screen image of a laptop showing a video call next to a warning icon, suggesting a deepfake scam attempt

How Attackers Use AI (and What Still Stops Them)

AI hasn’t handed criminals a new bag of tricks. It’s made the old ones faster and cheaper. That’s the short version of how attackers use AI right now, and it comes from the UK’s National Cyber Security Centre, which spent 2026 assessing what’s actually changed rather than what sounds scary in a headline.

How attackers use AI

The NCSC’s assessment is blunt about it: AI is almost certainly making cyber intrusion more effective and efficient, not inventing brand-new attack types. It speeds up things attackers already knew how to do. Scoping a target, writing convincing phishing messages, finding flaws in code, generating basic malware, sorting through stolen data. Same playbook, faster turnaround.

Two details matter more than the general trend. First, the time between a vulnerability being disclosed and someone exploiting it has shrunk to days, and AI is expected to shrink that further. An unpatched PC gives criminals less time to be found, but a much bigger opening if it stays open. Second, AI tools sold as a service are lowering the bar for less skilled criminals, letting novice scammers and hacktivists run opportunistic scams that used to require real technical skill. You’re not just facing elite hackers anymore. You’re facing a much larger pool of people with access to decent tools.

AI phishing and the voice on the other end

This is where deepfake scam and AI voice scam stories come from. Scammers now generate video and audio of real people, including public officials, and use it to add weight to a lie. A cloned voice saying it’s an emergency and money needs to move right now sounds far less far-fetched than it did five years ago.

The FBI’s Internet Crime Complaint Center documented a live version of this in 2026. Scammers built AI-made videos of a senior FBI official directing people to a fake IC3 website, dressed up to look almost identical to the real one. Victims who’d already lost money to one scam were told the FBI could help recover it, then were funneled through a spoofed form built to collect more of their information. The video wasn’t there to hack anything. It was there to make a lie feel official.

What still stops it

None of this requires a new category of defense. It requires doing the boring things consistently, because AI mostly attacks the gaps you already have.

  1. Patch promptly. Since the window between a flaw being disclosed and exploited is shrinking, delaying Windows updates and app updates by even a week or two matters more than it used to. See what a vulnerability actually is and why it’s worth closing fast.
  2. Verify anything urgent through a separate channel. If a call, video, or message pushes you toward money or personal details, hang up and contact the person or organization directly using a number or site you already know. IC3 recommends this exact step for spotting deepfake and impersonation scams: don’t click the link in the urgent message, go find the real one yourself.
  3. Watch for the intrusion itself. AI can write a better phishing email, but it still has to get onto your PC and do something once it’s there. That’s the point where network monitoring catches what a clever message missed. NetGuardian Pro watches network activity on your Windows PC and flags anything that looks wrong, with a plain suggestion for what to do next, whether the trigger was a phishing link or an old vulnerability nobody patched.

The FBI’s advice on spotting AI fakes is worth keeping in mind: distorted hands, mismatched shadows, voice lag, and unnatural pauses are still common tells, even in convincing fakes. Slow down before you react to anything that arrives with urgency attached. That single habit blocks more scams than any piece of software.

Sources

If it isn’t for you, there’s a 7-day refund window.

Related reading: AI makes scams more convincing. Read about how SIM swap fraud works and how to protect yourself.

About the author. Published by Joel Papenfus, founder of NetGuardian and creator of NetGuardian Pro. The sources this post draws on are linked in the post. Reviewed and checked by Joel on 1 October 2026.