Misinformation, Deepfakes and Checking Images

Edited photos, old videos with new captions and AI-generated images spread fast, especially during emergencies and big news events. A few quick checks can stop you being fooled, and stop you fooling your friends and family.

Know what you are up against

• Misinformation is false or misleading content shared without intent to deceive; disinformation is shared deliberately to deceive. Both spread the same way: through people who did not check.
• Much misleading content is not fully fake. It is real content with the context stripped away: an old photo from another country, a quote cut short, or a video clipped to change its meaning (Mike Caulfield, SIFT).
• Deepfakes are AI-generated or AI-altered videos, images or audio that make it look like someone said or did something they did not.
• Messages that create urgency ("Share before this is deleted", "Forward to everyone you know") are a warning sign in themselves.

Reverse image search in under a minute

• Save the image or take a screenshot. For a video, take a screenshot of a clear frame (PBS NewsHour).
• Upload it to a reverse image search tool such as Google image search (images.google.com, tap the camera icon) or TinEye (GIJN).
• On TinEye, sort by "oldest" to find the earliest appearance, or by "most changed" to find edited versions (GIJN).
• Look at when and where the image first appeared. If a "breaking news" photo was online years ago, it has been recycled.
• Apps that claim to detect manipulation can give a useful warning sign, but GIJN advises treating their results as a warning, not proof.

Spotting AI-generated and deepfake media

• MIT Media Lab's Detect Fakes project suggests paying close attention to the face: skin that looks too smooth or too wrinkled for the person's apparent age, odd shadows around the eyes and eyebrows, and glare on glasses that does not move naturally.
• Also check facial hair and moles, blinking (too much or too little) and whether lip movements match the words (MIT Media Lab).
• In images, look for objects that vanish or change between frames, physics that does not make sense and an unnaturally polished sheen (PBS NewsHour).
• PBS NewsHour warns that these visual clues "aren't as common as they once were". As tools improve, checking the source matters more than spotting glitches.
• AI detection tools can help but "are not always correct", and visible watermarks are easy to remove, so a missing watermark proves nothing (PBS NewsHour).

A quick checking routine

• Stop: notice your emotional reaction and do not share yet.
• Check who posted it first. Is it a known news organisation, an official account, or an anonymous account created recently?
• Search the claim with words like "fact check" and see whether reputable outlets are reporting it.
• Read the comments: others may already have found the original source or spotted problems (PBS NewsHour).
• During emergencies in the UAE, follow official government, police and civil defence channels rather than forwarded messages.

Why sharing responsibly matters in the UAE

• The UAE has a specific law, Federal Decree Law No. 34 of 2021 on Combatting Rumours and Cybercrimes, which came into force on 2 January 2022 (u.ae).
• The official portal says the law aims to combat the spread of rumours and fake news and lists offences and penalties for using information technology to spread false information.
• Forwarding something you have not checked can therefore carry legal as well as social consequences. If in doubt, do not share.
• If you have already shared something false, delete it and tell the people you sent it to. Correcting yourself is a sign of integrity.