For the sixteenth year in a row, the answer to "is the internet getting freer" is no. Freedom House's Freedom on the Net 2026, released on 22 October, found conditions deteriorating in 16 of the 38 countries it assesses and improving in only eight. It covers developments from June 2025 to May 2026, across a sample accounting for about 71 percent of the world's internet users.
The headline case is Iran, which recorded the year's largest decline and now sits with China and Myanmar as the worst environments for internet freedom. Its authorities imposed a nationwide shutdown lasting roughly 88 days — the longest in the report's history — and then rebuilt access so that only a privileged few could reach global platforms. Russia expanded shutdowns and VPN restrictions while keeping state-approved services reachable: an internet of permitted destinations.
Three mechanisms
First, isolation. Not censorship of individual pages but control over who is connected at all.
Second, age assurance. By June 2026, policymakers in 10 of the 38 countries had proposed, passed or implemented age-verification laws for social media or other non-pornographic content. Australia and Indonesia implemented bans for under-16s. Turkey's proposed framework would tie social media accounts to legal names through a government-run digital identity system. The report's objection is not that child protection is wrong. It is that this mechanism converts anonymity into an identity checkpoint for everyone — and the trail it creates is far more useful to a state than to a parent.
Third, AI-content rules. Governments pursued impersonation, harassment and child protection, but some used those laws to restrict online expression. Freedom House calls it plainly: measures that address real harm, weaponised.
Who moved
Georgia fell from Free to Partly Free as authorities continued to suppress online activism and speech perceived as critical of the ruling party. Jordan and Nigeria saw significant declines. Bangladesh improved the most for the second consecutive year, followed by the Gambia; Estonia remains the freest online environment. The United States' score was unchanged, with continued concern about surveillance and legal penalties for online activity: the TAKE IT DOWN Act's removal provisions took effect in May 2026, an important step against online sexual abuse, with open questions about safeguards against false reports.
"Authoritarian regimes are reengineering the internet into a system of control, deciding who can connect, what they can see, and whether they can be heard," said Kian Vesteinsson, the report's author and Freedom House's deputy director for research. "After 16 years of decline, democracies must lead the fight for a free and open internet."
What it asks for
Internet service providers should publish the shutdown orders they receive and commit to challenging disproportionate or unlawful ones. The US government should fund the Open Technology Fund, which builds circumvention tools for human rights defenders and journalists. Governments should push for emergency satellite connectivity at the ITU World Radiocommunication Conference in 2027. Funders should pay for research into age-assurance methods that genuinely preserve privacy.
That last recommendation is where the tension sits. The report asks for something that does not yet exist at scale: age verification without an identity registry. Anyone who wants to keep 16-year-olds out today needs faces, documents or accounts, and each of those produces precisely the data trail the report identifies as the danger. Child protection and anonymity are not inherently incompatible, but the technical answer is currently a research agenda, not a product category.
What the index does not show
Thirty-eight countries are not a picture of the world; the report covers about 71 percent of users and leaves much of Africa, Central Asia and Latin America out. Scores aggregate 21 indicators into a judgement, and the coverage window closes in May 2026 — anything since is uncounted. Read as a trend line, "16 years of decline" reflects a consistent methodology, not a fresh measurement of the same countries each year.
The common thread across the three mechanisms is still clear, and it has shifted: controlling content was yesterday's goal, controlling identity and reachability is today's. A firewall that gets circumvented is a political problem. An identity check that is a precondition for participation is a structural one.
