Malaysia: AI content labelling rules

The rules on labelling AI-generated content in Malaysia, what each asks of a file, and what one Kennmarq.com call adds.

In short

Images: No general rule found
Video: No general rule found
Audio: No general rule found
Text: No general rule found
3D: No general rule found
Rules: 1 for some sectors only.
In the API: markets=my

Updated 4 November 2026; sources checked 4 November 2026. Sources are government sites. This page explains what Kennmarq.com does; it isn't legal advice.

We found no general rule on labelling AI content here; the rules below apply to some sectors only.

Rules for specific sectors

Platforms

ONSA 2025 Risk Mitigation Code, para 4.2.4(d): In force since 1 June 2026. Prominent labels or markings on realistic generated or manipulated images, audio and video on the service.

In the API

Send markets=my on each call to POST /v1/mark: a call that lacks what Malaysia's rules need is refused before anything is counted, and each mark gets a receipt of what it carries for each law.

What a mark for Malaysia carries, by kind of file
File	Made with AI	A deep fake (`deepfake=true`)
Picture	Watermark and Content Credentials	Watermark and Content Credentials
Clip	Watermark and Content Credentials	Watermark and Content Credentials
Song	Watermark and Content Credentials	Refused for now: a song can't carry a visible label, and spoken notices for deep fakes aren't available yet: leave "deepfake" off for songs.

Mark a picture for Malaysia

curl https://api.kennmarq.com/v1/mark \
  -H "Authorization: Bearer $KENNMARQ_API_KEY" \
  -F file=@image.png \
  -F source=created \
  -F markets=my \
  -o image.marked.png

An AI coding agent gets the same answer from the MCP tool get_requirements (MCP server).

Malaysia on the map, and every country's rules side by side.