Using AI for Store Content and Service in Indonesia? The Rules Require Telling Buyers
Indonesia's Trade Minister Regulation 19/2026 requires sellers who use AI for content, product recommendations, promotion or service to inform consumers or label it. Here is how to decide where the label goes.Indonesia's Trade Minister Regulation (Permendag) 19 of 2026, signed on 8 June 2026, includes a provision on artificial intelligence in trading through electronic systems. According to Fortune Indonesia's summary of 11 June 2026, sellers who use AI to create content, recommend products, promote or provide service must inform consumers or label it. Marketplaces, for their part, must make sure AI-generated information can be accounted for.
For small sellers who have begun using AI for product photos, descriptions and chat replies, this changes one thing: buyers have a right to know. With year-end promotions approaching and content being produced faster and in greater volume, it is worth deciding now where that notice goes. This article walks through the AI disclosure rules Indonesia has put on record so far and what a small shop can sensibly do.
What is known and what is not
It is best to be straight about the limits. We know this duty from one media summary. That summary gives no article number, no required form for the label and no penalty. The regulation is written for trade through electronic systems, and the source does not explain how far it reaches into conversations outside a platform, such as direct chats with customers. So this article does not claim a label is mandatory everywhere. What we offer is sensible practice in the direction the rule points.
There is a wider development too. On 7 October 2026 Katadata reported that the presidential regulation (Perpres) on a National AI Roadmap was still being discussed and harmonised between ministries. Nezar Patria, Deputy Minister of Communication and Digital Affairs, said it was being discussed between ministries because its scope is broad. According to the report, the roadmap contains principles of responsible use, namely transparency, accountability and safety, and ministries and agencies will write more specific guidance for each sector.
That regulation has not been issued, so its principles are not legal duties yet. The direction is readable all the same: transparency and accountability. Sellers who build both habits now will have little to change later.
Where AI is used in your store
Start with a list. Most owners only realise how much AI their team uses once it is written down.
| Use | Example | Risk if not disclosed |
|---|---|---|
| Creating content | Product images made or altered by AI, descriptions, captions | Buyers assume the image is a photo of the real item |
| Recommending products | Automatically composed "right for you" suggestions | Buyers assume a person made the judgement |
| Promoting | Ad copy, testimonials, promotional video | Claims no human ever checked go live |
| Providing service | Automatic chat replies | Buyers assume they are talking to staff |
Three levels of use, three treatments
Not every use of AI weighs the same. The split below is our advice.
AI as a helper, a person finishes the work. Staff ask AI to tidy a description, then read and correct it. The final result is the staff member's responsibility. A general notice in the store profile is reasonable here.
AI makes something buyers see as a picture of the goods. AI-generated product images, models who do not exist, backgrounds altered so far that the item's colour shifts. Here the label should sit on the content itself, because buyers decide from what they see.
AI speaks directly to buyers. Automatic replies sent without anyone reading them. This most needs stating, at the start of the conversation, along with how to reach staff.
Plain wording for the notice
No legal language is needed. What is needed is a short sentence buyers understand. Some examples of our own that you can adapt:
- For images: "Illustration made with the help of AI. Photos of the actual item are on the next slide."
- For descriptions: "Description written with the help of AI and checked by our team."
- For automatic replies: "Hello, this is our store's automated assistant. Type ADMIN to be served by staff."
- For the store profile: "Some content and replies in this store are made with the help of AI and supervised by our team."
Two rules of use. First, never use an AI image as the only picture of the goods. Include real photos. Second, do not generate testimonials with AI. A label does not make an invented review honest.
Why staff-approved replies are easier to account for
The summary of the regulation uses the idea of AI-generated information that can be accounted for. In customer service that has a very practical meaning: if AI gives a wrong answer about price, stock or warranty, who carries it?
There are two ways to use AI for chat replies.
- AI answers on its own. Fast, but nobody reads the message before it goes. If the answer is wrong, the store finds out when the buyer complains.
- AI writes a draft, staff read it, correct it, then send. Slightly slower, but each message has a person who decided to send it.
Under the second way, the question "who is responsible" always has an answer. The buyer is dealing with staff helped by a tool, not a machine left alone. It is also easier to describe honestly: "our replies are drafted with the help of AI and checked by staff before sending."
For a small team, the second way suits questions whose answers carry money: price, shipping cost, availability, returns, warranty. Fully automatic replies are safer kept to things like opening hours and thank-you messages.
What to do this week
- List every use of AI in the store: who, for what, on which channel.
- Mark which ones buyers see directly.
- Put a notice in the store profile and on content whose images were made by AI.
- Set the types of question staff must read before an answer goes out.
- Recheck scheduled year-end promotion content: is there a claim no person has verified?
- Read the AI terms of each marketplace you sell on, since platforms may have their own labelling rules.
If you use Polynusa, customer replies can be drafted by AI and approved by staff before they are sent, so each message still has someone accountable.
Note: this article summarises news reports and is not legal advice. Confirm the duties for your business against the text of Regulation 19/2026, the platforms you sell on, or a consultant.
Sources
- Fortune Indonesia, 11 June 2026: What the latest e-commerce trade regulation contains
- Katadata, 7 October 2026: Government prepares sector-specific AI rules
FAQ
Must online sellers in Indonesia disclose AI use?
According to Fortune Indonesia's summary of Trade Minister Regulation 19/2026, sellers who use AI to create content, recommend products, promote or provide service must inform consumers or label it.
What form must the AI label take?
The sources we read do not state the form of the label, an article number or a penalty. The wording in this article is suggested practice, not an official format.
Is the national AI roadmap regulation in force?
No. Deputy Minister Nezar Patria said on 6 October 2026 that the presidential regulation was still under discussion between ministries, as Katadata reported.