Deepfakes and AI Content: How to Spot It and Protect Yourself

AI can now make convincing images, voices and video of almost anyone. Here's how to stay sceptical without becoming paranoid.

• Images of events that never happened.
• Voices cloned from a short recording.
• Video of people saying or doing things they didn't.
• Text that imitates someone's style.

• Odd hands, teeth, text, reflections or backgrounds in images.
• Lip movements slightly out of sync, or unnatural blinking and lighting in video.
• A voice that's unusually flat or has odd breathing and pauses.
• Emotional urgency ("act now, don't tell anyone").
• No original source or an unfamiliar account posting it.

Detection by eye is getting harder, so don't rely on it alone.

1. Check the source: who first posted it? Do reputable outlets confirm it?
2. Reverse-image search suspicious pictures.
3. Look for content credentials (C2PA) or watermarks where available.
4. Verify requests out-of-band: if "your boss" or "a family member" asks for money, call them on a known number.
5. Agree a family or team code word for emergencies.
6. Pause before sharing.

• Save evidence, report to the platform, and tell the person affected.
• Many places have laws against non-consensual synthetic imagery and impersonation. Report to the authorities.

• Disclose AI-generated content where appropriate.
• Keep originals and records to prove authenticity.
• Don't use real people's likeness or voice without permission.

Trust the process, not your eyes: verify sources and use a second channel for anything involving money or reputation.