SINGAPORE, 7 AUG 2026 — In the first half of 2022, Malaysia asked Google to take down 46 pieces of content. In the first half of 2025 — the most recent period Google has published — it asked for 27,484. Across those same three years Vietnam's demands fell from 7,470 to 1,645.
The two countries are not doing the same thing at different volumes. They are doing different things, and the reason codes in Google's own data say so.
What we counted
Google publishes every government content-removal request it receives as a downloadable dataset, broken down by country, requesting body, product, stated reason, and what Google did about it. We took the bulk file and read the ASEAN rows across all thirteen half-year periods it covers.
One limitation shapes everything below. The most recent period in the dataset is January to June 2025, because Google reports with a substantial lag. This is the current state of the public record, not the current state of the world.
| Six months to | Malaysia | Indonesia | Viet Nam | Thailand | Singapore |
|---|---|---|---|---|---|
| Jun 2022 | 46 | 567 | 7,470 | 1,097 | 265 |
| Dec 2022 | 116 | 1,995 | 4,172 | 714 | 301 |
| Jun 2023 | 164 | 1,908 | 11,527 | 699 | 43 |
| Dec 2023 | 112 | 7,376 | 5,330 | 264 | 30 |
| Jun 2024 | 430 | 11,624 | 6,416 | 467 | 510 |
| Dec 2024 | 5,054 | 14,566 | 4,431 | 299 | 175 |
| Jun 2025 | 27,484 | 20,190 | 1,645 | 224 | 159 |
Items of content requested for removal by government bodies, per half-year, from Google's published bulk dataset. Items, not requests: one request can cover many items. Google product coverage only, and Google's own reason attribution. Extracted by RECATOOLS on 7 August 2026; January to June 2025 is the most recent period in the file.
Two regimes, not one trend
Sorted by stated reason, the region splits cleanly.
In Malaysia, 26,914 of the 27,484 items were filed under regulated goods and services. In Indonesia the same category accounts for 19,084 of 20,190. In Vietnam it is government criticism that dominates, at 1,362 of 1,645, and in Thailand the same category covers 134 of 224.
Google defines regulated goods and services as claims of infringement of local laws covering, among other things, the illegal sale, trade or advertising of pharmaceuticals, alcohol, tobacco, fireworks, weapons, gambling, prostitution and health or medical devices. Government criticism it defines as claims of criticism of government policy or politicians in their official capacity.
So the growth in ASEAN's takedown volume is not a growth in speech policing. On this data, speech-based demands are falling where they were once most common, while market-policing demands have risen steeply where they were once rare.
The category is broad and the dataset does not decompose it, so we cannot say from this file what the Malaysian and Indonesian filings actually concern. Gambling is one of the listed items in Google's definition, and this region has a documented illegal online gambling economy, but the data does not attribute the filings and neither do we.
Malaysia's rise is a programme, not a filing
A jump from 430 items to 27,484 in twelve months invites the suspicion that one bulk submission distorted the series. It did not.
In Malaysia's most recent period, the Information and Communications Authority asked for 25,604 YouTube removals for regulated goods and services, an item count spread across 941 separate requests. That is roughly five filings every day of the six months, weekends included, from one agency against one product.
Malaysia is now the largest source of removal demands in ASEAN by item count, ahead of Indonesia, which is by far the region's most populous country.
One product, one agency
The concentration is worth stating plainly, because it narrows the story.
Of Malaysia's 27,484 items, 27,416 concern YouTube. Web Search accounts for 34 and Gmail for 11. The Information and Communications Authority filed 26,127 of them, with government officials accounting for a further 1,325. This is one agency, one platform, one reason code.
Indonesia is close but not identical, and the difference matters. Its Information and Communications Authority filed 20,148 of the country's 20,190 items, but they split across YouTube at 13,726 and Web Search at 6,222. Indonesia is asking for search de-listing at meaningful scale; Malaysia is essentially not.
The outcomes reflect that distinction. Indonesia's search-directed filings under regulated goods and services ran to 6,216 items with 2,200 acted on — 35 per cent, against 91 per cent for the same country's YouTube filings under the same reason. Removing a video a platform hosts and de-listing a page it merely indexes are different asks, and Google treats them differently.
Vietnam is concentrated too, at 1,514 items from its Information and Communications Authority and 129 from police, almost entirely on YouTube. Singapore is the outlier in shape rather than scale: its 159 items come mostly from bodies the dataset labels only as "Other", with police accounting for 29.
Google does not treat the two regimes alike
The responses, not the demands, are the most revealing figures, and they diverge inside a single country.
| Country and stated reason | Items requested | Items removed | Compliance |
|---|---|---|---|
| Malaysia — regulated goods and services | 26,914 | 24,941 | 93% |
| Malaysia — government criticism | 88 | 11 | 12% |
| Indonesia — regulated goods and services | 19,084 | 13,772 | 72% |
| Indonesia — hate speech | 778 | 709 | 91% |
| Viet Nam — government criticism | 1,362 | 833 | 61% |
| Thailand — government criticism | 134 | 129 | 96% |
| Singapore — regulated goods and services | 107 | 91 | 85% |
January to June 2025. Items removed combines Google's legal and policy removals. Compliance percentages are RECATOOLS arithmetic over the two columns shown. Rows with small item counts move sharply on a handful of decisions and should not be read as stable rates.
Malaysia is the clearest case. Google acted on 93 per cent of the items it was asked to remove for regulated goods and services, and on 12 per cent of those filed as government criticism. Same requesting state, same six months, a near eight-fold difference in outcome depending on what the request was for.
Vietnam's government-criticism demands are met 61 per cent of the time, on a base roughly fifteen times larger than Malaysia's. Thailand's are met 96 per cent of the time, but on 134 items, which is small enough that a few decisions move the figure a long way.
Across all reasons, the Philippines has the region's lowest compliance rate at 26 per cent, on 168 items.
Reading these as a measure of how much Google resists a given government would be a mistake. Compliance tracks the legal basis of the demand as much as its subject. A request citing a specific local statute on unlawful advertising is a different proposition from one asserting that criticism of a minister should come down, and the numbers behave accordingly.
What this changes for anyone operating in the region
For a platform or publisher, the compliance surface in Southeast Asia is shifting from a speech problem to an advertising and marketplace problem, which requires different teams and different tooling.
Twenty-seven thousand items in six months from one Malaysian agency, against one product, is an operational load that a legal-review workflow built for defamation notices will not absorb. It implies automated detection on the government side and, on the receiving side, a process that can act at similar speed.
For users, the effect is quieter and harder to see. Content removed under a regulated-goods heading attracts no scrutiny, because almost nobody objects to illegal-gambling advertising being taken down. That is precisely why the category deserves the attention: it is the fastest-growing removal reason in the region, it has the highest compliance rate, and it is the least examined.
What to watch
Two things to watch for in the next release.
The first is whether Malaysia's volume holds. A single half-year at 27,484 could be a campaign against a specific problem that ends. Two consecutive periods at that level would make it structural, and would make Malaysia one of the higher-volume removal requesters anywhere.
The second is whether Vietnam's decline continues. Falling from 11,527 to 1,645 in two years is a substantial change in behaviour by a state that has not become more permissive in other respects, and this dataset cannot say whether the demands stopped or simply moved to channels Google does not report.