On 28 October 2026, Deputy Communications Minister Teo Nie Ching gave the Dewan Rakyat the most detailed account yet of how Malaysia polices false online content. Between 1 January 2023 and 15 October 2026, the Malaysian Communications and Multimedia Commission (MCMC) sent 65,680 removal requests to service providers. Of these, 56,304 (86%) were taken down, according to The Star's report of the written answer. Over the same period MCMC investigated 546 false-content cases under Section 233 of the Communications and Multimedia Act 1998 (Act 588).
The case for the current system
The strongest argument for this approach is practical. False content about health, public finance and scams can cause harm in hours, and criminal trials take years. A request that a platform review a post against its own community guidelines is faster and less coercive than prosecution. The numbers also suggest MCMC is not using its harshest tool by default. Most investigations ended without a charge, and the takedown volume dwarfs the prosecution volume. If the choice is between asking platforms to act and jailing speakers, the request route is the milder one.
What the numbers actually show
The Section 233 outcomes sum exactly to 546, per the same Star report:
- 390 cases (about 71%) needed no further action.
- 74 cases ended in warning letters.
- 34 were compounded, for RM1.72 million in total.
- 19 went to court. Twelve have been decided, with fines totalling RM75,500.
- 29 remain under investigation.
Three features stand out. First, roughly seven in ten investigations were closed with nothing, which means the investigation itself was the main consequence for most of those people. Second, only 19 cases (about 3.5%) reached a court, where a judge tests the evidence. Third, compounds raised about 23 times more money than court fines. Dividing the totals gives roughly RM50,600 per compound against roughly RM6,300 per decided court case. These are our own averages from the reported totals, not official figures. They suggest that an offer to settle can cost more than going to trial.
That inversion is the policy concern. A compound is a payment to avoid prosecution, so it never tests whether the content was actually false or fell within the statute. A person who settles has an incentive to pay quickly, whatever the merits. The data cannot tell us how many of the 34 payers would have been acquitted, because no court heard those cases.
Removal without a stated legal test
The takedown side is less transparent still. Teo said removals rested on "assessments of complaints received, compliance with platform community guidelines and applicable laws." That sentence combines three different standards. A complaint is an allegation. A community guideline is a private contract term. Only "applicable laws" is a public legal test, and the answer as reported names no specific provision for the 65,680 requests. The Star's account cites only Section 233, which governs the 546 investigations.
This matters because the 86% figure is easy to misread. It could mean platforms agreed that the content was false. It could equally mean platforms comply with a regulator's request rather than argue with it. The published numbers cannot separate the two. They also do not break down which of the 9,376 refused requests (14%) were declined and why, or how many removed items were later restored. Without that, nobody can judge whether the 86% reflects accurate flagging or deference.
The statute itself points toward restraint. Section 3(3) of Act 588 states that nothing in the Act is to be construed as permitting the censorship of the Internet, as the Sinar Project notes. The same source records that Section 233 is widely used alongside the Penal Code and Sedition Act 1948 and that courts have often treated its restrictions as constitutional. Civil society groups have long urged reform because the section's wording is broad. The 2025 amendments, which took effect on 11 February 2025 after gazettement on 7 February, changed other provisions, according to Malay Mail. The minister said Section 233 itself was retained. The breadth that critics object to therefore remains in force.
A proportionate reform agenda
None of this requires abandoning enforcement against fraud and genuine incitement. It requires measuring the system properly. Four steps would make the existing powers more defensible without weakening them:
- Publish the legal basis of each takedown request. Aggregate counts should be split by statute, by community-guideline referral, and by complaint category such as scams, gambling or political speech.
- Report refusals and reversals. The 14% of requests that were not honoured, and any content reinstated, are the only evidence that the process can say no.
- Report the content type behind compounds. If settlements involve political criticism rather than fraud, the public should know.
- Notify the affected speaker. A person whose post is removed on a regulator's request should be told why and given a route to challenge it.
Malaysia's deputy minister closed by saying the government would keep strengthening efforts "through legislation and appropriate enforcement action." A government that prefers removal requests to prosecutions has a reasonable case, but it should be able to show the evidence. At present the public can see that content is vanishing at scale and that money is changing hands in settlements. It cannot see the legal tests applied in either process. Transparency on those two points would do more for public trust than another round of statistics on how much was removed.
The MCMC's published text of Act 588 is the place to start for readers who want to check the provisions themselves. The figures for the next reporting period will show whether the pattern holds: low court volume, high compound income, and a takedown rate that no one outside MCMC can audit.