Spotting Deepfake Content in Photos and Video
Launch library · evergreen read
Deepfake technology can now convincingly alter faces and voices in video footage, making it appear that someone said or did something they never actually did in reality. As the technology continues to improve, spotting these fakes reliably by eye alone becomes progressively harder, though a few useful signals still help.
Unnatural blinking patterns, lighting that does not quite match properly between a face and its surrounding background, and audio that is slightly out of sync with lip movement are all worth noticing, alongside a source or overall context that seems unusually convenient for the specific claim being made.
Considering carefully where a video actually came from matters just as much as how it visually looks. A shocking clip from an unfamiliar account, with no coverage at all from established, genuinely verifiable news sources, deserves real scepticism even if the footage itself happens to look entirely realistic.
Before sharing anything sensational further, a quick search to see whether reputable outlets are reporting on the same event helps avoid spreading a fabrication, deepfake or otherwise, any further than it has already unfortunately travelled across social media that particular day and night, largely unchecked.