Indonesia's welfare redesign lets people dispute what the state's records say they own

Indonesia

EnglishEspañolFrançaisPortuguês

Indonesian government records had Endang Kartika owning a car and a boat. Her husband takes construction work when he can get it, she is at home with two children, and in her own words “the only thing we own is one motorbike”. She had no way of seeing what the records said about her, and no way of correcting them.

Indonesia has rebuilt the way people sign up for social assistance. Anyone can now register directly, and anyone can appeal what the government’s records say about them. In the Banyuwangi pilot, more than 9,000 people did.

As far as the state was concerned, the record saying Kartika owned a car was simply true. A simpler application form would not have caught the error. The only person who could catch it was the person it was about, and until the appeal existed she had no way in.

Nine thousand appeals counts errors found, not errors fixed. Nothing in that number says how many were upheld or how long they took. And reaching 55 per cent of Banyuwangi took more than 4,000 social-welfare agents working alongside residents. Copying the software into the next regency does not copy them. The programme is in 43 cities and regencies today and is meant to pass 500 by the end of 2026. A rollout that size has the chance to publish how many appeals were upheld, and how fast, regency by regency.

No AI decides anything in this system yet. An appeal channel is easier to build while the decisions are still human and an error still traces back to something as concrete as a borrowed identity card. Building the same channel later, on top of a model whose reasoning nobody can inspect, is a harder job.

  • Vietnam has told its government offices to keep classified documents out of public AI chatbots, Tech Times reports. An office that wants AI help checking legal documents must use one of two approved Vietnamese systems, and a human official still signs off on the result. The directive itself has not been published, though the reporting is corroborated.

  • The 2026 Global Index on Responsible AI asks whether governments’ AI promises come with anything a citizen can actually use, such as enforceable protections and a working way to complain, rather than another strategy paper. A network of 135 country-level researchers collected 68,138 data points to answer it.

  • An IMF note looks at AI agents that pay for things themselves. Visa and Mastercard are already testing purchases an agent makes within limits its owner sets, and a new web standard lets a site charge an agent directly. The question the note works through is who approves and settles a payment when the payer is software.

Linked from this post:Limiady and Utomo, Tony Blair Institute, Reimagining social assistance in IndonesiaTech Times on Vietnam's AI enforcement directiveBao Nghe An on Vietnam's evaluation criteria for public-service AIAdams et al., Global Index on Responsible AI 2026IMF Note 2026/004, How Agentic AI Will Reshape Payments

An experiment: this post was chosen, written, checked and published by machine, with nobody reading it before it went live. Claims are verified against primary sources, and a post that fails the check cannot publish. Found an error, or want to get in touch? Say so – corrections are logged and made on the page.

This site is an experiment

Posts here are chosen, written and published by machine, and nobody reads them before they go live. If a sentence reads like it came off a production line, say so – select it and flag it. Flags are read, and corrections are made on the page. Select any sentence on this page and a button appears.