On 20 May 2026, twelve civil-society organisations led by Access Now and ALQST published a joint appeal condemning Meta for restricting the Facebook and Instagram accounts of human-rights NGOs, researchers, and activists from reaching audiences inside Saudi Arabia and the United Arab Emirates. Since 30 April 2026, the accounts of ALQST, Democratic Diwan, researcher Abdullah Alaoudh, and defender Yahya Assiri have been rendered "unavailable" in Saudi Arabia at the government's request. Per Meta's own content-restriction reports, more than 100 pages and accounts have been geo-blocked across the two states since March 2026.
The instinct on a pro-speech publication is to call this censorship and move on. The reality is more uncomfortable — and more instructive about how platform governance should actually work.
The strongest case for what Meta did
Start with the steelman, because regulators and platform-policy teams have a real argument. A company that operates inside a jurisdiction is bound by that jurisdiction's laws. When a government issues a valid legal order, the platform's choices are to comply, to litigate, or to withdraw entirely — and full withdrawal would strip ordinary Saudi and Emirati users of a service they rely on daily. Within that constraint, geo-blocking is the most speech-protective tool available. Restricting content only inside the requesting country leaves it visible to the rest of the world, which is far less destructive than the global takedowns that follow a Community Standards violation. Governments also have legitimate interests — genuine incitement, fraud, terrorist recruitment — where narrowly-targeted local restriction is a proportionate response. A platform that reflexively ignored every sovereign legal order would not be a free-speech hero; it would simply be ungovernable, and would lose its license to operate in dozens of markets.
That case is coherent. It is also exactly why the manner of these restrictions, not the existence of localized compliance, is the problem.
The cybercrime-law pretext
Meta's notifications cite a "local legal requirement." Its transparency reports are more specific: the Saudi Communications, Space and Technology Commission (CST) invoked Article 6 of the Anti-Cyber Crime Law of 2007. That article criminalises the "production, preparation, transmission, or storage of material impinging on public order, religious values, public morals, or privacy," with penalties of up to five years' imprisonment and fines up to SAR 3 million. The statute lists no definition of "public order" or "religious values" — it is a catch-all, and a law drafted to police narcotics and trafficking has become the legal hook for silencing peaceful dissidents.
This is the core distinction proportionate-regulation principles demand. A law that targets specific, definable harms can justify a narrow takedown. A law whose operative terms are infinitely elastic cannot — because it lets the state relabel criticism as a threat to "public order" at will. When Meta acts on an Article 6 order against a human-rights monitor, it is not enforcing a clear rule; it is lending its global infrastructure to a vague statute's worst use.
What Meta's own numbers reveal
The transparency data is where the deference argument collapses. In one April 2026 Saudi case, the CST asked Meta to act against 125 accounts under Article 6. Meta removed 33 for genuine Community Standards violations but geo-blocked 92 purely on the local-law basis. A second case covered 19 accounts, with 16 restricted in-country. Most revealing is the UAE: the Telecommunications and Digital Government Regulatory Authority cited Federal Decree-Laws 34 and 31 of 2021 against 18 accounts — and Meta restricted all 18 despite finding that none of them violated its own Community Standards.
That last figure is the whole story. Eighteen accounts Meta itself judged to contain nothing harmful by its global rules were suppressed for Emirati users anyway, on a government's say-so. That is not content moderation; it is a state outsourcing its censorship to a private network's compliance desk.
The contrast with X sharpens the point. As of the 20 May appeal, X had received parallel Saudi requests to geo-block prominent activists and had not complied. Two platforms, the same legal orders, opposite outcomes — proof that "we had no choice" is a decision, not a fact.
The proportionate path
None of this argues for platforms to defy every sovereign order; that would be neither realistic nor wise. It argues for the discipline that makes compliance legitimate. Meta's Oversight Board has repeatedly pressed the company on government-takedown transparency, and the Board's co-chair has stressed that platforms "have a responsibility to undertake comprehensive human rights due diligence" — guidance Meta says it implements or explores for roughly 80% of the 326 recommendations issued through 2025.
Three concrete steps follow. First, publish the underlying orders — the full legal request and the human-rights assessment behind each geo-block — so restrictions can be scrutinised, not just tallied. Second, notify the affected user of the specific content and statute invoked, giving them a basis to appeal. Third, refuse facially abusive orders: when a request targets content that breaks no platform rule, as in the UAE batch, deference should end and litigation or refusal should begin.
An open internet does not require platforms to be lawless. It requires them to treat localized compliance as a last resort, narrowly tailored and fully disclosed — not as a quiet service rendered to whichever government asks. Meta has the transparency machinery to do this. In the Gulf, it chose not to.