How to Spot a Deepfake: A Practical Guide for Everyone
Slug: how-to-spot-a-deepfakePillar: Technology > Online SafetyKeyword: how to tell if a video is a deepfakeExcerpt: Deepfakes are increasingly convincing, but they still leave detectable clues. Learn the practical checks to spot fake video, audio and images.
What Is a Deepfake and Why Does It Matter?
A deepfake is a piece of media — video, audio, image, or text — where AI has been used to replace or fabricate a person's face, voice, or actions convincingly. In 2026, the best ones can fool forensic tools. But the majority still have detectable flaws if you know what to look for. Deepfakes are used in scams, disinformation, and harassment. The UK's Online Safety Act now makes sharing non-consensual intimate deepfakes a criminal offence.
How to Spot a Deepfake Video
Look at the hairline, ear boundaries, and the area where the face meets the neck. Blurring, flickering, or inconsistent skin tone in these areas suggests manipulation. Watch the eyes closely — corneal reflections are often inconsistent in deepfakes. Teeth are another tell: AI still struggles to render realistic teeth. Check whether the lip movements precisely match the audio — a slight desync on hard consonants like "p," "b," and "m" is a reliable indicator.
How to Spot a Deepfake Audio (Voice Clone)
Listen for: unnatural pauses between words, robotic rhythm on multi-syllable words, or emotion that doesn't quite match the content. Background ambience is often absent — real recordings have room noise; synthetic audio is frequently too clean. If someone calls you urgently requesting money or sensitive information, always verify through a separate channel — call them back on a known number.
How to Spot a Deepfake Image
Look at hands and fingers — AI still regularly produces extra fingers, merged fingers, or anatomically impossible positions. Earrings, glasses frames, and jewellery are frequently asymmetric or distorted. Text within images (signs, labels) is often garbled nonsense. Run suspicious images through a reverse image search (Google Images or TinEye).
Verification Tools to Use
Hive Moderation (hivemoderation.com) offers a free image and video deepfake detector. FotoForensics (fotoforensics.com) analyses image metadata and compression artefacts. The InVID/WeVerify browser extension is widely used by journalists to verify video authenticity. No single tool is definitive — use them alongside visual checks and source verification.
What to Do If You Suspect a Deepfake
Don't share it. If it involves financial fraud or threatens safety, report it to Action Fraud (actionfraud.police.uk). If it's non-consensual intimate imagery, report to the platform and to the Internet Watch Foundation (IWF).
Frequently Asked Questions
Are deepfake detectors reliable?
They're useful but imperfect — current detectors achieve 80–95% accuracy in controlled tests but can be fooled by high-quality deepfakes.
Can I tell by metadata if a video is fake?
Sometimes. AI-generated videos often lack standard camera metadata. Tools like ExifTool can check this.
Is it illegal to create deepfakes in the UK?
Creating non-consensual intimate deepfakes is now illegal under the Online Safety Act 2023.
How do I protect my own likeness from being deepfaked?
Limiting publicly available photos and videos of yourself reduces the training material available. Set social profiles to private and watermark professional headshots.
More online safety guides at eight2infinity.com/technology
