TikTok's Q1 2026 enforcement figures for the Middle East and North Africa give an unusually concrete picture of how moderation works in Saudi Arabia. Between January and March 2026, the platform removed 2,811,302 videos in the Kingdom, with a 98.2% proactive removal rate, and restored 150,916 videos after successful appeals. Across eight MENA markets the total was 15,519,240 removals. These are TikTok's own numbers, not an independent audit, but they are enough to ask a useful policy question: what law is this moderation operating under, and does it give users predictable rules?
The strongest case for the current model
The argument for aggressive proactive moderation is serious. Short-video platforms carry spam, scams, graphic violence and material harmful to minors at enormous volume, and no user can be expected to report it all. A 98.2% proactive rate means that almost all removals happened before a user flagged the video. That reduces exposure to harm. It also means fewer disputes about who reported whom. A regulator could reasonably say that a platform acting quickly and appealing transparently is what a proportionate regime looks like. The appeal data supports part of that: 150,916 reinstated videos is roughly 5.4% of the removal count, a visible sign that mistakes are caught and corrected. (That is a ratio of two quarterly totals, not a measured error rate. TikTok's report does not say how many appeals were filed.)
The legal backdrop is not neutral
Platform rules are not the only rules in play. Saudi Arabia's Anti-Cyber Crime Law, Royal Decree M/17, adopted on 26 March 2007, criminalises the production, preparation, transmission or storage of material that impinges on public order, religious values, public morals and privacy. Article 6 of the official English text sets the penalty at imprisonment of up to five years, a fine of up to SAR 3 million, or both.
The concern is not that Saudi Arabia has a cybercrime statute. Every jurisdiction does. It is that the categories in Article 6 are defined by reference to broad social values, not by identifiable harms such as fraud, credible threats or child exploitation. 'Public order' and 'public morals' have no fixed content. A user cannot tell in advance where lawful commentary ends and an offence begins.
The provision is used in practice. In late 2025, Gulf News reported that the General Authority of Media Regulation referred six people to the Public Prosecution over online content deemed to incite public opinion, citing Article 6(1) and its penalties of up to five years and SAR 3 million.
Why the combination matters
Put the two facts together. A platform that serves Saudi users has a commercial and legal incentive to avoid content that could fall under Article 6. Because the statute is open-textured, the safest compliance strategy is to over-remove. TikTok's report does not say what share of the 2.8 million Saudi removals involved content that might touch Article 6. It also does not break out which policy categories drove them, so we should not claim that over-removal is happening. What can be said is that the legal structure rewards caution, and the published data cannot show whether that caution is excessive.
The appeals figure is therefore informative but incomplete. A reinstatement rate says the platform's internal process corrects some errors. It says nothing about speech a user never bothered to appeal, because they feared drawing attention to a post in a jurisdiction with criminal penalties. In a market where the downside of contesting a takedown can include scrutiny by prosecutors, low appeal volume is not evidence of low error.
The quarter-to-quarter pattern is also worth watching. TikTok's previous report put Saudi removals at 2,903,914 in Q4 2025, with a 99.9% proactive rate and 146,314 videos restored. Q1 2026 is slightly lower on removals and the proactive rate, and slightly higher on restorations. Two quarters cannot establish a trend, and the earlier report does not say that its restorations followed appeals, so the two restoration figures are not strictly comparable.
A proportionate path forward
A pro-innovation, pro-speech position does not require abolishing moderation or denying that harmful content exists. It requires three things.
- Narrow the statutory text. Replace value-laden categories with defined harms and a requirement of intent or demonstrable risk. Narrower offences also help platforms, because they can write rules that map to the law.
- Publish category-level data by country. Aggregate removal totals cannot be tested. Platforms should break down Saudi removals by policy reason, the share reviewed by humans, and appeal outcomes, so researchers can see whether legal risk shapes enforcement.
- Separate platform enforcement from state referral. Users should be told when a removal results from a government request versus a platform's own rule. That transparency would reveal how much of the 2.8 million, if any, reflects legal pressure rather than community guidelines.
The numbers TikTok released are a start, and companies deserve credit for publishing them. But the volume of automated enforcement is only one half of the picture. The other half is a criminal provision whose reach cannot be read off its text. Until that gap is closed, a 98.2% proactive rate tells us the machine is efficient, not whether it is aimed at the right targets.