AI Regulation Tracker  /  Draft rule

Indonesia drafts AI ethics rule with three risk tiers, but frames it without sanctions

Komdigi has sent two draft Presidential Regulations on artificial intelligence to the law ministry for finalization, one of which would sort AI systems into three risk classes. Neither is signed or in force, and the government says the ethics rule would carry no penalties of its own.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 7, 2026  ·  Last updated July 7, 2026

Two draft rules would move Indonesia's AI governance from a 2023 soft ethics circular to a binding Presidential Regulation, but they are still drafts, and the government is signaling that the flagship one will carry no penalties of its own.

The Ministry of Communication and Digital Affairs, known as Komdigi, has advanced two draft Presidential Regulations, or Peraturan Presiden. One addresses AI ethics. The other sets a national AI roadmap. According to reporting current as of April 2026, both drafts have moved from Komdigi to the law ministry and state secretariat for finalization, the stage before a Presidential Regulation is sent to the president for signature. Neither has been signed. Neither is in force. No Presidential Regulation number has been issued, and reporting professionals should not attach one.

What the ethics draft would do

The ethics draft is the one to watch. It is reported to introduce a three-tier risk categorization for AI systems, with harm-mitigation duties keyed to each risk class. That structure echoes the European Union's AI Act, which sorts systems by risk and attaches heavier obligations to higher-risk uses. If signed, the ethics Perpres would convert Indonesia's 2023 soft ethics circular, a non-binding guidance document, into a binding governance framework. It would also serve as the parent instrument for derivative Ministerial Regulations, including a separately reported rule on labeling AI-generated content.

The three-tier structure is significant because of what it replaces. Indonesia's existing AI governance rests on a 2023 ethics circular, a soft-law document that sets expectations without creating enforceable obligations. Moving from a circular to a Presidential Regulation would harden that guidance into a binding framework and, importantly, create a legal anchor for the more specific rules Komdigi wants to issue underneath it. The most concrete of those is a reported Ministerial Regulation on labeling and watermarking AI-generated content, which the government has said would take effect only after the AI Perpres is signed. In other words, the ethics draft is the keystone: several coming duties are waiting on it.

The second draft, the national AI roadmap, sets strategic direction rather than duties. Its planning period has been cited inconsistently in Indonesian coverage, appearing as both 2026 to 2030 and 2025 to 2029. Neither period is confirmed here, and both are reported as they appear in the source coverage. A roadmap of this kind typically allocates priorities, funding, and institutional roles rather than imposing obligations on private firms, so its direct compliance weight is low even after adoption.

The catch: framed "without sanctions"

The nuance that most reporting outside Indonesia has missed is enforcement. The government has publicly described the ethics Perpres as "tanpa sanksi," meaning without sanctions. On that framing, the ethics rule would set standards and duties but would not impose penalties directly. Enforcement would instead fall to the existing Electronic Information and Transactions Law, known as the ITE Law, which already handles content and platform liability in Indonesia and routes penalties through its own provisions.

That matters for how firms should read the tiers. A risk-tier framework with no penalty attached is a governance benchmark and a signal of direction, not a fine schedule. The teeth, to the extent there are any, sit in a separate statute. It also means the practical consequences of a signed ethics rule may arrive indirectly, through the derivative Ministerial Regulations it authorizes, rather than through the Perpres itself. This attribution, that the ethics rule is framed without sanctions and that enforcement stays with the ITE Law, rests on government statements reported by Komdigi and Indonesian outlets including Katadata and Hukumonline, not on a promulgated text a reader can yet inspect.

What it does not do

The drafts do not change any obligation today. Because they are unsigned, they create no duty a firm can breach, no deadline to meet, and no sanction to avoid. There is no promulgated Presidential Regulation number to cite, and any specific enactment reference circulating in secondary coverage should be treated as unverified until Komdigi's official legal database publishes the signed text. The three tiers, the harm-mitigation duties, and the labeling rule that would flow from the framework all depend on a signature that has not happened.

For a US reader, the significance is directional. Indonesia is a large market, and a risk-tiered ethics framework there would, once signed, bind foreign AI providers by risk class when their systems reach Indonesian users, in the same pattern the EU AI Act established. It also confirms that the EU's risk-tier model is becoming a template that other jurisdictions copy, which shapes the baseline any global AI deployer will eventually design to. For now, though, the correct posture is monitoring, not compliance. The framework is coming. It is not here.

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.