Deepfake technology is increasingly accessible — and alongside that, legal liability is tightening. This article is not just about using AI to detect deepfakes; it is primarily about the obligation to disclose.
IMPORTANT LEGAL WARNING: Creating or distributing deepfake content WITHOUT disclosure is illegal in many countries, including Vietnam. Disclosure is MANDATORY when using deepfake — violations can result in criminal or civil prosecution. This article is for educational purposes only and does not constitute legal advice. If you are unsure about your legal obligations, consult a lawyer.
What is deepfake and why should creators care?
Deepfake is image, video, or audio content that is created or edited by AI to depict a person saying or doing something they did not actually say or do. In a content creation context, this includes:
- Videos of a public figure "endorsing" a product they never reviewed
- AI voice clones reading a script the person never recorded
- Face-swapped video replacing the original performer
- Real people's images composited into fictional contexts
Deepfake disclosure obligation — non-negotiable
Disclosure is MANDATORY when you use AI to create or edit content depicting a real person in a way that could cause confusion. Major platforms including YouTube, TikTok, and Meta all have policies requiring labelling of AI-generated content. Laws in many countries (increasingly enforced in Vietnam) clearly define this obligation.
How to disclose correctly:
- Clear labelling in the description: "This video uses AI/deepfake technology"
- Add a watermark or on-screen overlay on deepfake segments
- Follow each platform's AI labelling policy for every platform you post to
- Do not use another person's likeness or voice without their explicit consent
AI deepfake detection tools
Note: Deepfake detection tools are assistants — AI is a draft, you still need to review. No tool achieves 100% accuracy, especially against high-quality deepfakes.
Common tools
- Microsoft Video Authenticator: Analyses video frame-by-frame to detect signs of AI editing
- Sensity AI: Deepfake detection service for enterprises and media organisations
- FakeCatcher (Intel): Uses blood flow analysis to detect synthetic video
- Deepware Scanner: Free tool for basic screening
Warning signs to look for
- Lip movement out of sync with audio
- Unnatural blinking or no blinking at all
- Lighting on the face inconsistent with the background
- Unnatural hairline or ear edges
- Skin texture that is too smooth or shows AI artefacts
Creator responsibility
If you use AI to create content involving a real person — whether their face, voice, or actions — you are responsible for:
- Obtaining consent from that person (if living) or their family/legal representative
- Clearly disclosing in the content and metadata
- Not using deepfake to spread misinformation, defame, or defraud
Save reference videos on deepfake law and disclosure via Klypio YouTube downloader or @KlypioBot.
Also see: product placement disclosure guide, copyright and fair use for creators.