Indonesia's Supreme Court Secretary has adopted media guidelines that allow AI as a support tool but keep court officials responsible for what is published

Key takeaways

• The instrument is a Supreme Court Secretary decree on media and social-media management, dated 9 October 2026 as reported. We have not opened it.
• As reported, AI can be used as a support tool but does not replace the authority and responsibility of the authorised officials.
• As reported, AI-assisted content has to be checked, verified and approved before publication.
• The report gives no effective date, no sanctions and no labelling duty, and it does not extend the guidelines to judicial decision-making.

What is the Supreme Court Secretary's media decree, and is it binding?

It is a decision of the Secretary of the Supreme Court, Decision No. 14106/SEK/SK.HM1.1/X/2026 dated 9 October 2026, which Dandapala reports adopts guidelines for managing mass media and social media in the Supreme Court and the courts under it. The acting Secretary, Sobandi, discussed the guidelines at an online socialisation session on 20 October 2026, organised by the Bureau of Law and Public Relations.

The report does not say how binding the guidelines are, when they take effect, or what follows from non-compliance, and we have not opened the decree. In our reading it is an internal administrative instrument, but that is the desk's characterisation and not the source's. The report states the aim as professional, effective, integrated and responsible media management, with attention to information security and public trust in the courts. For the neighbouring professional-body instrument, see our piece on the Peradi generative AI guidelines for advocates, which is a separate document from a bar body.

What does the decree say about AI?

The report says the guidelines regulate the use of artificial intelligence in media management. In its words: "Teknologi AI dapat digunakan sebagai alat bantu, tetapi tidak menggantikan kewenangan dan tanggung jawab petugas yang berwenang." (our translation: AI technology can be used as a support tool but does not replace the authority and responsibility of the authorised officials).

On publication, the report says: "Konten yang dihasilkan dengan bantuan AI harus diperiksa, diverifikasi, dan disetujui sebelum dipublikasikan." (our translation: content produced with AI assistance has to be checked, verified and approved before it is published). It adds that managers remain responsible for the accuracy, security and content of the information given to the public.

The report also says speed of content production should not override fact-checking, personal-data protection, copyright and the principle of judicial independence. It reports no named tools, no disclosure or labelling requirement for AI-assisted posts, and no penalty. Those absences are limits of the report, not confirmed gaps in the decree.

What else in the guidelines touches AI-assisted publishing?

The same report describes a general publishing workflow that AI-assisted content would sit inside. Each work unit is to have a management team with clear roles from planning through production, editing, verification, publication and evaluation, and every item goes through verification and approval before it is posted. Information is to be accurate, drawn from official information, and free of personal opinion that could suggest taking a side.

For cases that draw public attention, the report says the information published on a court's media follows the substance of a press release already published by the Supreme Court, which courts can send through the Bureau of Law and Public Relations. Official social-media accounts are to use institutional rather than personal email and contact numbers, with two-factor authentication or a similar feature. The report does not link these points to AI use, and we do not.

What should court-media teams, vendors and foreign counsel do now?

Treat this as a narrow signal about court communications, not about AI in judging. The report supports a few practical steps, each limited to what it describes.

• Court-media teams: identify where AI is used in drafting, translating or editing posts, and name the person who checks and the person who approves before anything is published.
• Vendors: be ready to show how a product keeps a human approver in the loop. The report names personal data and copyright as concerns but sets out no technical standard.
• Foreign counsel: obtain the decree text from the Supreme Court before relying on any detail here, and check whether other court-administration instruments cover AI use outside media.

Related court-AI items elsewhere in the region include the Philippines Supreme Court AI judiciary framework, and the separate instruments listed under related briefings below.

What are the key dates and parties?

What we did not verify

What we opened: the Dandapala report published 20 October 2026, read in full on 2026-10-25. We read the full article text. Dandapala is the judiciary's own outlet, so this is the institution's account, not independent reporting.

What we did not open: the decree itself, any official Supreme Court publication of the guidelines, the separate Dandapala items listed as related reading on that page, and any implementing instruction for subordinate courts. We do not know the decree's legal basis, effective date, annexes or sanctions, if any.

What we refuse to claim: that the guidelines are legally binding on anyone, that they apply to AI in judicial decision-making, that they require labelling or disclosure of AI use, or that any court has been sanctioned under them.

Related briefings

• AI Regulation News hub
• Asia-Pacific AI regulation tracker
• AI laws for attorneys
• Peradi generative AI guidelines for Indonesian advocates
• Philippines Supreme Court AI judiciary framework
• India Supreme Court draft rules on AI in courts
• Nepal judiciary ICT policy on AI in judgment writing