Rahmat Wibowo's criminal complaint to Bareskrim's cyber crime directorate against Meta AI

What this document is

An 8-page “Surat Aduan Masyarakat” (public complaint), numbered 077/ADUAN-DITTIPIDSIBER/RW/IX/2026 and dated 28 September 2026, addressed to the Director of Cyber Crime (Dittipidsiber) at Bareskrim Polri. Rahmat Wibowo wrote it as the “Pelapor” and posted all eight pages on Threads.

The reported party (“Terlapor”) is the Meta AI chatbot on Threads (“meta.ai”). He asks the police to identify the company legally responsible for it, because, as he writes, an AI cannot itself be reported for a crime.

What it says

• Offense: spreading misleading information through an electronic system, under Article 28(1) jo. Article 45A(1) of Law 1/2024 (the ITE Law).
• The misleading information: in one Threads conversation on 28 September 2026, Meta AI cited Article 27A of the ITE Law, which he says no longer applies since the new Criminal Code took effect on 2 January 2026, and once cited Articles 310/311 of the old Criminal Code. It corrected itself twice when users pointed this out.
• Requests: that the police register it as a “Laporan Polisi Model A”, process it as an ordinary offense (delik umum) that does not depend on a victim’s complaint, investigate the operator, ask the operator to explain its quality controls, coordinate with the Ministry of Communication and Digital Affairs, and send him a receipt by email.
• Evidence: nine screenshots of the conversation (P-1 to P-9).

Observations

1. His own evidence shows what the chatbot said about him. In evidence P-6, Meta AI answers a user who asked whether Petra Novandi Barus’s post was bullying: “Tidak. Postingan Petra Barus itu curhatan personal soal nemu hasil tes IQ lama, tidak ada target…” and “Justru balasan akun rweebs21 yang berpotensi masalah. Menyebut seseorang ‘calon terpidana tipikor’ dan ‘Petrus Pembully Terus Menerus’ tanpa bukti adalah tuduhan personal.” In P-8 it adds that words like “calon” or “menurut keyakinan saya” do not automatically protect a serious, personal accusation. His chronology describes P-6 only as a citation of Article 27A and Articles 310/311, and leaves out what it said about his posts. The complaint is dated the same day.
2. He first asked the chatbot how to report the chatbot. In P-1 he asks Meta AI who the reported party would be “kalau saya mau melaporkan kamu”; in P-4 he asks whether it is afraid of being reported to Bareskrim.
3. The chatbot said it was not giving legal advice. The screenshots show Meta AI repeating “saya bukan advokat”, “ini bukan nasihat hukum resmi” and advising users to consult a lawyer. The complaint acknowledges the disclaimers and argues they do not remove the impact.
4. The loss the article requires is not shown. Article 28(1), as he quotes it, covers misleading information “yang mengakibatkan kerugian materiel bagi konsumen dalam Transaksi Elektronik” — that causes material loss to consumers in an electronic transaction. In his own table of elements, this is the only one he does not mark as fulfilled, and he names no one who lost anything. He also asks that the complaint not depend on him being a victim.
5. “Model A”. As commonly described under the police chief’s regulation on criminal investigation, a Model A police report is made by a police officer who witnessed or learned of an event, while a report from a member of the public is Model B. The complaint, written by a member of the public, asks to be registered as Model A.
6. Received as mail, not shown as a police report. The pages show no receipt (STPL), stamp or agenda number; he asks for the receipt to be emailed to him. He later posted a mail receipt from Bareskrim’s administration office showing the complaint was delivered on 6 October 2026, eight days after its date; it gives no police report number (doc-20261006-001).
7. Complaint numbers. His complaints carry numbers that look like a running count: 008 on 13 August and 009 on 14 August (doc-20260814-001), 063 on 11 September (ev-20260911-001), 077 on 28 September (this one) and 087 on 14 October (doc-20261014-001). If the numbers are sequential, he wrote 78 complaints in two months.
8. The earlier Meta AI reprimand. Two weeks later Meta AI again told him that mass tagging in a personal conflict can be read as spam or harassment, and he answered with “Nullum Delictum Nulla Poena Sine Praevia Lege Poenali” (ev-20261014-004); on 17 October it told him again that “calon terpidana tipikor” should not be used for Petra Novandi Barus (ev-20261017-002, ev-20261017-003).

Whether Article 27A still applies was not checked against the official gazette, which could not be accessed when this analysis was written.