Malaysian Media Used AI to Illustrate the National Flag. Most Got It Wrong.
What happened
In April 2025, multiple Malaysian organisations, including major newspapers, published images of the Jalur Gemilang generated by AI tools. The images were wrong. Stars were misshapen or miscounted, the crescent was drawn incorrectly, and proportions were off in ways immediately obvious to anyone who had grown up looking at the flag. The errors were not marginal. They were published anyway.
The public reaction was swift and severe. Police reports were filed. Calls for boycotts circulated against the outlets responsible. Malaysia's King, Sultan Ibrahim, described the errors as unacceptable, and the prime minister weighed in on the question of editorial judgment. For a flag that carries the weight of national sovereignty, Islamic identity, and the country's multiracial history, a generated misrepresentation was not treated as a harmless technical glitch. It was treated as a failure of respect.
What happened is straightforward: AI image tools are trained on datasets that often contain incomplete or inaccurate representations of national flags, particularly those of countries that appear less frequently in training corpora. The Jalur Gemilang's specific geometry, the number and placement of its stars, the precise arc of its crescent, is not the kind of detail a generative model reconstructs reliably without high-quality reference data. The organisations that used these tools either did not verify the output before publication or did not know what to verify against.
The controversy deepened because the errors did not land in neutral territory. Commentators noted that flag disputes in Malaysia carry racial and political dimensions, and that incorrect depictions of an Islamic symbol on the national flag could be read, selectively or genuinely, as disrespect toward Malay and Muslim identity. What began as a quality-control failure compounded into something that exacerbated existing tensions and was available to be weaponised by those who found it useful to do so.
The gap the incident exposes is not primarily about AI accuracy. It is about the absence of a verification step between what a system produces and what an organisation publishes under its own name. A provable record of what a system generated, who reviewed it, and what standard the output was checked against would have caught this before it reached print or screen. Without that record, the question of accountability after the fact collapses into competing claims about who knew what, and the damage is already done.
Reported impact
- Affected parties
- Not publicly disclosed
- Harm type
- Not publicly disclosed
- Scale
- Not publicly disclosed
- Financial impact
- Not publicly disclosed
- Regulatory action
- Not publicly disclosed
Classification
Relevant governance controls
Governance control mapping is not available for this record.
- No controls mapped
Not publicly disclosed
Control mapping is analytical. It does not state that any control would have prevented the incident.
Sources and evidence
This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.