If you have ever asked an AI tool to make a product photo, a promo poster, or an Instagram caption, this week's news from Indonesia's Ministry of Communication and Digital Affairs (Komdigi) is worth a few minutes. Not because there is anything to panic about. The rule is still under review and has no effective date. But its direction is clear, and getting ready costs very little if you start now.

What was actually reported

On Saturday, 17 October 2026, Deputy Minister Nezar Patria said Komdigi is reviewing a ministerial regulation on labelling AI-generated content. The reasoning: AI audio, video, and images look more and more like the real thing, which makes them harder to tell apart and raises the risk of fraud. According to InfoPublik and Bloomberg Technoz, the companies expected to label authenticity are AI developers.

A day later, Deputy Minister Angga Raka Prabowo tied the review to AI-made videos, images, and audio about forest and land fires. He also said several platforms already attach an "AI content" notice, as quoted by RCTI+.

This did not come out of nowhere. On 26 January 2026, Digital Ecosystem Director General Edwin Hidayat Abdullah told parliament's Commission I that a draft ministerial regulation would require generative AI content to carry a watermark or label, and that offending content could be taken down, according to ANTARA and Investortrust. The October coverage mentions neither penalties nor an issue date.

Timeline of the AI content label story: 26 January 2026 a draft regulation raised in parliament, 17 October the ministry still reviewing, 18 October some platforms already labelling, as of 21 October no effective date

So, are shops and cafés affected?

From the reports we read, the obligation points at AI developers and platforms. January's coverage names platforms, October's names AI developers, so the exact scope is genuinely unclear. None of the reports names a small business that uses AI for its own promotion as a party that must comply.

But the effect can still reach you by another route. The AI tool you use may add a mark automatically. A platform may attach an "AI content" notice to your post, as some already do. And if an asset is questioned or taken down, the first question is almost certainly the same: who made this, with what, and when? You want to be able to answer that in a minute.

A label is not evidence

People tend to treat a label and a record as the same thing. They are not. A label is the mark your audience sees on the asset. A tool or platform adds it, and it can vanish as soon as an image is cropped or screenshotted. A provenance record is yours: who made it, with which tool, when, and which part the AI helped with.

Comparison of the label on the asset, which people see and which can disappear, with the provenance record, which you keep and can show when asked

Picture a bakery (an illustration, not a real case) that used AI to generate the background of a promo poster. One day a platform flags the poster. The bakery with a record answers quickly: made by the shop admin on this date, with this tool, only the background is AI, and the original file is saved. The bakery without one is left guessing and digging through old chats.

There is a practical reason too. AI tools change quickly. A year from now you may be on a different tool, and the memory of who used what will have faded. A record written today saves you from guessing later.

Seven columns are enough

The record does not need to be complicated. One spreadsheet, one row per asset, filled in when the asset is finished, not when someone asks.

Seven columns for each AI-made asset: asset and where it runs, person responsible, tool and date, AI-made part, original file, label applied, reuse

The "person responsible" column is the one people skip most, and the most useful. If one named person always approves AI-made assets before they go live, ownership is clear and nobody passes the buck.

What you can do this week

One afternoon is enough. No budget needed.

  1. List every asset that uses AI: product photos, posters, captions, promo videos, chat replies.
  2. Name one person who decides whether an AI-made asset can go live.
  3. Start the record, beginning with what is live today.
  4. Check label settings in the tools and platforms you use. If there is an option to mark AI content, switch it on.
  5. Follow official Komdigi announcements: who is covered, what the label looks like, and the penalties.
Five steps for one afternoon: list AI assets, one person responsible, start a record per asset, check labels in tools and platforms, watch official Komdigi news

What you do not need to do

You do not need to stop using AI. You do not need to stamp a big "AI" on every asset before the rule exists, and you do not need to buy a special tool. The label format, who is covered, and the penalties only become clear once the ministerial regulation is issued. Until then, a tidy record is enough to keep you and your customers at ease.

Closing

If you would like to talk through how to record your assets and tidy up your workflow, the XETUP team is happy to be a sounding board. Message us on WhatsApp at +62 823 1499 5005 and tell us your situation first.