Deep Fake AI

What it is.

"Deepfake" refers to AI-generated audio, images, or video made to look or sound like a real person - convincingly enough to fool most people at a glance or a listen. The tools to make one have gotten dramatically cheaper, faster, and more convincing, and they no longer require any special skill.

Why it matters now.

What used to require a Hollywood budget can now be done by anyone with a laptop. That's fueling a sharp rise in:

• Political disinformation - fabricated video or audio of public figures saying things they never said, timed to spread before it can be debunked
• Voice-cloning scams - a call that sounds exactly like a family member in trouble, asking for money or gift cards, often using audio clipped from social media or a voicemail

How scammers actually get your voice to clone it

A short clip is all it takes - 10 to 15 seconds of clear audio from a social media video, podcast appearance, or even a voicemail greeting is often enough for current tools to produce a convincing clone. They can also get it live: the old "Hello? Can you hear me?" call, originally known as a trick to record you saying "yes" for billing fraud, works just as well today to harvest your natural speech pattern for cloning - even a casual "Who is this?" said in your own cadence can be enough. The fix is the same either way: don't engage. Don't say "who is this" or anything else - just hang up. If a call like this becomes a pattern, consider letting unknown numbers go to voicemail instead of answering live.

• Harassment and impersonation - fake images or video used to threaten, embarrass, or extort someone
• Business fraud - fake video calls or voice messages impersonating an executive to authorize a wire transfer

How to spot one.

No single tell is reliable on its own, but watch for a combination of:

• Unnatural blinking, or eyes that don't quite track normally
• Lighting or shadows that don't match the rest of the scene
• Audio that's slightly out of sync with lip movement, or a voice with an odd cadence, flat emotion, or strange pauses
• Skin, hair, or jewelry that looks slightly too smooth or shifts oddly between frames
• A sense of urgency in the message - deepfake scams rely on you reacting before you think

Tools that can help analyze suspected deepfakes

Worth knowing: every detector is playing catch-up against whatever generation of AI model made the fake, so treat any result as one signal, not proof - the same as the manual red flags above.

InVID/WeVerify (VeraAI) - a free Chrome/Firefox extension built specifically for journalists, fact-checkers, and OSINT investigators by an EU-funded research consortium. Reverse image/frame search, metadata analysis, and manipulation checks in one toolkit - this is the actual tool working journalists and groups like Bellingcat use, worth leading with over any random AI-detector startup.

Other free/consumer options: AI or Not (quick image/audio checks), Hive Moderation (free browser demo), Deepware Scanner (free video-focused checker), and McAfee's Deepfake Detector (bundled with some McAfee products). Enterprise-grade tools like Reality Defender, Sensity AI, Intel's FakeCatcher, and Pindrop (audio-specific) are what platforms, banks, and newsrooms license rather than something an individual signs up for.

What to do about it.

• If someone contacts you claiming to be a relative, official, or coworker in an urgent situation asking for money or sensitive info, hang up and call them back on a number you already have - not one they gave you. If they claim to be in custody, call the authority they claim has them to verify.
• Agree on a family "safe word" in advance for real emergencies.
• Don't trust a video call alone as proof of identity; ask a question only the real person would know. Safe word!
• Reverse-image or reverse-video search suspicious media before sharing it.
• Report suspected deepfake scams to the FTC (reportfraud.ftc.gov) and, for identity theft or fraud attempts, the FBI's IC3 (ic3.gov/PSA).

The FBI has issued specific public warnings about AI voice-cloning and deepfake-enabled fraud targeting phone users - worth a read if you want more detail: FBI Warning For All iPhone, Android Users.