AI Regulation Tracker / Draft rule
Indonesia is drafting a rule to make platforms label AI-generated content
Indonesia's digital ministry told parliament it is writing a regulation to require labels or watermarks on AI-generated content. Nothing is in force yet, and the timeline depends on a separate AI Presidential Regulation that has not been signed.
Indonesia's Ministry of Communication and Digital, known as Komdigi, told parliament on January 26, 2026 that it is drafting a rule to require labeling or watermarking of AI-generated content. The announcement came during a working meeting with Komisi I of the DPR, Indonesia's House of Representatives. Nothing is in force. What exists today is a stated intention and a draft in early development, not a binding obligation.
The signal matters because of where it points, not because of what it requires right now. According to reporting by the state news agency ANTARA and by CNBC Indonesia, the Director General of the Digital Ecosystem, Edwin Hidayat Abdullah, described the planned rule as a derivative instrument that would sit beneath a broader AI Presidential Regulation, or Perpres. Minister Meutya Hafid has backed the effort. The practical consequence is that the labeling rule cannot take effect on its own. It waits on the Perpres, which itself remains a draft awaiting the President's signature.
What the rule would require
As described by Komdigi, the planned Ministerial Regulation, or Permen, would bind all electronic system operators, referred to in Indonesian law as PSE, to apply a label or watermark to AI-generated content. The stated purpose is transparency: helping users tell human-made content from machine-generated content and curbing misuse such as deceptive synthetic media. That framing places Indonesia alongside a growing set of jurisdictions moving toward provenance and disclosure duties for AI output.
How it would be enforced
Enforcement would not arrive as a standalone fine. Komdigi has indicated that non-compliant content would be subject to takedown, with penalties routed through Indonesia's Electronic Information and Transactions Law, known as the UU ITE. That is the existing statutory backstop Indonesian regulators already use for prohibited online content. In other words, the labeling duty would borrow the ITE Law's takedown machinery rather than create a new fines regime. For a US reader, this is the important structural point: the obligation, if it lands, would be enforced by removal of content, not by a published monetary penalty schedule. It also means the practical risk is loss of reach and distribution in Indonesia rather than a fine, which changes how a compliance team would weigh the exposure.
What it does not do yet
It is essential to be precise about the current state. There is no regulation number. There is no effective date. There is no published text. The rule is a draft in development, and it is legally inert until the AI Presidential Regulation is signed. A firm reading only headlines could easily overstate this as a live mandate. It is not one. Until the AI Presidential Regulation is signed and the derivative Permen is issued with an operative date, there is no labeling obligation a company can violate, and no watermark specification a platform is required to meet.
A related correction is worth stating plainly, because Indonesian press coverage has conflated two different countries. A widely circulated "AI labeling rule effective September 1, 2025" is China's rule, issued by the Cyberspace Administration of China and tied to the GB 45438-2025 standard, not Indonesia's. Indonesia has no in-force mandatory AI-content-labeling law. The only Indonesian instrument in this space is the Komdigi draft described here, together with the unsigned AI Perpres it derives from. Anyone assessing Indonesian exposure should not attribute China's September 2025 measure to Indonesia.
The cross-border angle
For US firms and professionals, the reach is the reason to pay attention. Indonesia regulates electronic system operators on a territorial-effect basis, so a rule binding "all PSE" would reach foreign platforms that serve Indonesian users, not only companies incorporated in Indonesia. If the Permen is adopted, a US platform, publisher, or professional distributing AI-generated content to an Indonesian audience would owe the same labeling duty as a domestic operator. That said, the trigger has not been pulled. The correct posture today is monitoring, because the duty is coming into view but has not become law.
Frequently Asked Questions
Does Indonesia currently require AI-generated content to be labeled?
No. As of July 7, 2026, Indonesia has no in-force mandatory AI-content-labeling law. Komdigi has announced it is drafting a Ministerial Regulation, but there is no regulation number, no published text, and no effective date.
Who would the rule apply to if it is adopted?
All electronic system operators (PSE), including platforms and businesses that publish AI-generated content to Indonesian users. It would reach foreign platforms serving the Indonesian market, not only Indonesian companies.
Is this the same as the AI labeling rule that took effect in September 2025?
No. The rule effective September 1, 2025 is China's, issued by the Cyberspace Administration of China under the GB 45438-2025 standard. Indonesian press has conflated the two. Indonesia's version is still only a draft.
When would Indonesia's rule take effect?
No date has been announced. The Ministerial Regulation is a derivative of a separate AI Presidential Regulation that has not been signed, so it cannot take effect until that Perpres is finalized.
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.