From 12 October 2026, users in Azerbaijan reported that TikTok, Instagram and X had stopped working across several internet providers. TechRadar reported that NetBlocks confirmed the blocks on 13 October. The disruption followed protests against new school dress rules that keep girls who wear the hijab out of class. Proton VPN's account says the government approved the rules on 21 September, barring clothing that conflicts with the "secular nature of education". According to 301.am, 37 people were arrested in Baku, and criminal proceedings were opened under Article 233 of the Criminal Code.
The evidence on the blocks themselves is not uniform, and that matters. 301.am cites Al Jazeera reporting that OONI and NetBlocks data did not support claims of a widespread internet shutdown, which leaves room for platform-specific or provider-level blocking. No Azerbaijani authority has confirmed or explained any restriction. The facts we can say with confidence are narrow: users report three platforms failing, monitors disagree on the scope, and the state has been silent.
The strongest case for the government
A fair reading starts with what a government might say. Street protests can turn violent. Mobilisation calls spread quickly on social platforms, and a state responsible for public order and for school safety may think a short, targeted pause is less harmful than a crowd confrontation. Proton VPN's account quotes an anonymous Baku activist saying protest calls were circulating on Instagram and TikTok, many from abroad. Governments also have a real interest in curbing incitement. Secular-education policy is a legitimate area for democratic debate, and reasonable people disagree about dress rules in schools.
That argument has some force. It still fails on three counts.
Why the blocks fail a proportionality test
The measure is not targeted. A platform block cuts off every user on that platform, not the handful who might incite violence. Parents trying to reach schools, small businesses selling through Instagram, and journalists sharing footage all lose access along with organisers. International human-rights standards, including Article 19 of the International Covenant on Civil and Political Rights, ask that restrictions on expression be necessary and proportionate. A blanket block on three major platforms is the opposite of narrowly tailored.
It is not transparent. Restrictions imposed without any announced legal basis cannot be challenged, appealed or reviewed. Silence is the central problem here. Without a published order, citizens cannot tell whether a court, a regulator or an ISP made the decision, and no one can be held accountable for overreach. A government confident in its rationale would state it.
It is ineffective, and it misfires. Users report reaching the platforms through VPNs, so motivated organisers keep communicating. The people who lose access are the ordinary users who do not use circumvention tools. Blocking also signals that a policy cannot withstand open debate, which tends to feed the grievance driving the protests. Arrests under criminal law already give the state a way to act against violence. Taking the platforms offline adds cost without adding security.
This is not an isolated pattern
The context shows an established toolkit rather than a one-off. Freedom House's Freedom on the Net 2025 rates Azerbaijan "Not Free" at 34/100, with 9 out of 40 for violations of user rights. It records that authorities routinely throttle connections and block websites, including critical media. In January 2025, access was reportedly throttled during a protest in Imishli, and residents were reportedly forced to delete social media posts about the incident. In February 2025, an appellate court in Ganja ordered a news site blocked at the request of the Media Development Agency, which cited a registration issue, while the site's editor linked the case to his refusal to remove critical articles.
Globally, the trend is rising. Access Now's KeepItOn coalition counted at least 313 shutdowns in 52 countries in 2025, which it describes as the highest number since 2016. It defines shutdowns to include throttling and platform blocks, not only full network cuts. That definition is why a three-platform block belongs in the same category as a national blackout: both disrupt the means of speech, only at different scales.
What proportionate policy looks like
If Azerbaijan's goal is public order, the measures that fit are well known. The state can prosecute specific unlawful acts, such as actual incitement to violence, through courts with evidence and legal review. It can request removal of specific posts from platforms under a published process. It can engage with the school communities affected by the dress rules instead of waiting for grievances to spill into the street. None of these requires cutting off millions of users.
The platforms bear some responsibility as well. Meta, TikTok and X have not publicly explained the disruption. Companies that earn revenue in a market should publish clear information about government demands they receive and push back on unlawful ones where they can. Transparency from both sides would let the public learn who decided what.
There is also an innovation cost that rarely enters these debates. Digital businesses, creators and online sellers depend on predictable access. A market where platforms can be switched off without notice is a market where investment is discounted. That is a concrete reason, beyond rights, for governments to treat blocking as a last resort and not a first reflex.
The takeaway
Whatever the scale of the technical measures, the policy problem is clear. A dispute over school dress rules is a question for education policy and the courts. Throttling or blocking social platforms treats it as an information-control problem, and it harms ordinary users more than the people it targets. If the restrictions are in place, Azerbaijan should say so, name the legal basis, set an end date and restore access. If they are not government-ordered, it should say that too. Silence is the least defensible position.