Reporting · Data investigation
2016: A Coup On Internet Rights In Turkey
How emergency measures slowly became permanent digital control
Imagine something significant is happening in the country (an earthquake, an election, an explosion). You reach for your phone to find out more. Twitter won’t load, Instagram keeps refreshing, and WhatsApp messages don’t get sent. You open a VPN, and that doesn’t work either…
Today, this is the reality for millions of users in Turkey: the internet stops working the moment it is most needed. It has become so common that many people living in Turkey now just assume something has happened as soon as social media becomes inaccessible.
This experience is the result of almost two decades of progressive state-level internet control, shaped by a structural shift in the year 2016. Before 2016, our data shows that restrictions were sporadic and reactive; afterward, they became systematic, politically framed, and routine.
As of 2023, İFÖD counted +953,415 blocked domains, +260,000 blocked URLs, and +67,100 blocked tweets. In 2024, Roblox, Discord, and Wattpad were blocked by court order. These remain blocked, as we enter 2026.
The Coup Attempt That Changed Everything
On the night of July 15th, 2016, a group of the Turkish army allegedly linked to Fethullah Gülen attempted and failed at a coup d’état against the government and President Recep Tayyip Erdogan. That same night, while tanks were rolling through Turkish streets, Facebook, Twitter, and YouTube went dark for two hours. Organizations like AccessNow and Turkey Blocks reported intentional throttling of these sites.
Just a month prior to the coup, an “internet kill-switch” regulation prepared under the authority of the government’s telecom regulator (BTK) and published in the Official Gazette on 11 June 2016 had come to effect. The 10th clause of the aforementioned regulation gave BTK the regulatory power to “partially or entirely” suspend internet access due to war and/or national security, without seeking ministerial approval.
MADDE 10 – Aynı Yönetmeliğin 21 inci maddesinin başlığı aşağıdaki şekilde değiştirilmiş ve aynı maddeye aşağıdaki fıkra eklenmiştir. “Hizmet sunumunun engellenmesi ve işletmenin askıya alınması” “(2) Kurum; savaş, genel seferberlik ve benzeri durumlarda, kamu güvenliği ve ulusal savunma için gerekli görüldüğü takdirde, sınırlı veya sınırsız bir süreyle işletmecinin işletme faaliyetlerinin tamamını veya bir kısmını askıya alabilir ve doğrudan doğruya şebekeyi işletebilir. Bu durumda geçecek süre, yetkilendirme süresinin sonuna eklenir.”
Resmî Gazete (Official Gazette), 11 June 2016 in Turkish
Additional non-comprehensive data shows at least 14 throttling and full-shutdown incidents that year alone. All of these incidents coincided with significant moments of political tension including security operations, protests, and the attempted coup itself. So, was 2016 just a temporary escalation tied to extraordinary circumstances or a transitional moment into internet control as a governing tool?
Before we attempt to answer that question, we would like to share our analysis of Freedom House’s and Reporters Without Borders’ scores of freedom (political rights and civil liberties) and freedom of press to give you an understanding of the context that shapes our analysis.
Freedom House Scores Turkey
Freedom Index from Freedom House showing the decline in civil liberties and political rights in Turkey.
These changes largely coincide and correlate with the coup attempt and a suspicious increase in internet connectivity issues across the country.
- Political rights
- Civil liberties
- Total
Source: Freedom House: Freedom in the World
Show the data
| Year | Political rights | Civil liberties | Total |
|---|---|---|---|
| 2006 | 28 | 37 | 65 |
| 2007 | 28 | 37 | 65 |
| 2008 | 29 | 37 | 66 |
| 2009 | 29 | 36 | 65 |
| 2010 | 27 | 36 | 63 |
| 2011 | 27 | 36 | 63 |
| 2012 | 28 | 35 | 63 |
| 2013 | 28 | 33 | 61 |
| 2014 | 28 | 32 | 60 |
| 2015 | 26 | 29 | 55 |
| 2016 | 24 | 29 | 53 |
| 2017 | 18 | 20 | 38 |
| 2018 | 16 | 16 | 32 |
| 2019 | 15 | 16 | 31 |
| 2020 | 16 | 16 | 32 |
| 2021 | 16 | 16 | 32 |
| 2022 | 16 | 16 | 32 |
| 2023 | 16 | 16 | 32 |
| 2024 | 17 | 16 | 33 |
| 2025 | 17 | 16 | 33 |
Press Freedom Scores of Turkey
Reporters Without Borders' (RSF) Press Freedom Score
The index presented shows the reported press freedom in Turkey which is a result of factors such as the freedom of journalists to report on any topic they choose but also on ownership relations of news corporations and policians.
Interpretation: It is important to understand that the Press Freedom Index is an INDEX. This means, the resulting score is relative to all other countries covered by reporters without borders AND previous years' values for a single country. To understand the methodology in detail, visit RSF's methodology page.
No score is published for 2011.
Source: Reporters Without Borders
Show the data
| Year | Score |
|---|---|
| 2002 | 33.5 |
| 2003 | 35 |
| 2004 | 37.2 |
| 2005 | 25 |
| 2006 | 25 |
| 2007 | 31.2 |
| 2008 | 22.8 |
| 2009 | 38.2 |
| 2010 | 49.2 |
| 2012 | 70 |
| 2013 | 53.4 |
| 2014 | 54.1 |
| 2015 | 55.8 |
| 2016 | 49.2 |
| 2017 | 47.0 |
| 2018 | 46.5 |
| 2019 | 47.2 |
| 2020 | 50.0 |
| 2021 | 50.2 |
| 2022 | 41.2 |
| 2023 | 34.0 |
| 2024 | 31.6 |
| 2025 | 29.4 |
The Freedom House and RSF data shows a clear regression of freedom and press freedom over the years that is particularly steep around 2016. While the press freedom score recovers slightly around 2021 (before falling again), the civil liberties and political rights scores never truly recover after 2016.
Our Methodology
We analyzed 7 independent datasets covering nearly two decades of internet shutdowns, content blocking, and political freedoms. These include records of internet shutdowns and throttling incidents from AccessNow’s #KeepItOn project (2013-2024), reports and tweets from NetBlocks and Turkey Blocks (2016-2023), network measurement data from the Open Observatory of Network Interference (2016-2025), longitudinal indicators of political and press freedom from Freedom House and Reporters Without Borders, and news records of shutdown/throttling incidents (2007-2014).
Data cleaning and validation protocol Google Docs · switch to the “final” tab
We aggregated shutdowns, content-blocking, throttling, and any other reported restrictions by year and manually cross-checked across multiple sources to verify dates, scope, and context. Then, we categorized them by type of restrictions (full shutdowns, throttling, platform-specific blocking, etc.) and by context (elections, protests, security operations, or other politically significant events). Additionally, we analyzed OONI web connectivity measurements for Turkey between 2016 and 2025. We focused on categories relevant to political communication, including news outlets, social networks, communication tools, and circumvention technologies. To account for changes in measurement volume over time, we calculated the proportion of failed or anomalous tests relative to total measurements, allowing for year-to-year comparison.
Because connectivity tests and reporting density on restriction incidents widely vary over time, especially before 2016, our analysis does not claim to cover every instance of internet restriction. What we tried to do was identify long-term patterns, shifts in frequency, and changes in framing/justification.
Our Findings
Establishing a Pattern – The NetBlocks and AccessNow data
Over time, our data revealed a change in the frequency and the character of Turkey’s internet restrictions.
2016 as a Peak Year for Documented Internet Restrictions in Turkey
The number of reported internet restrictions peaks sharply in 2016, coinciding with the failed coup attempt and the introduction of emergency regulatory powers.
The data reflects reported incidents and does not represent a comprehensive record of all restrictions.
Source: Access Now, NetBlocks / TurkeyBlocks, Freedom House, and publicly reported incidents.
Show the data
| Year | Incidents |
|---|---|
| 2007 | 1 |
| 2008 | 1 |
| 2009 | 1 |
| 2010 | 5 |
| 2011 | 3 |
| 2012 | 7 |
| 2013 | 4 |
| 2014 | 4 |
| 2015 | 0 |
| 2016 | 12 |
| 2017 | 3 |
| 2018 | 4 |
| 2019 | 2 |
| 2020 | 1 |
| 2021 | 0 |
| 2022 | 1 |
| 2023 | 2 |
| 2024 | 2 |
| 2025 | 2 |
From Content Blocking to Network-Level Disruption (2007–2025)
This chart displays the yearly distribution of documented internet restriction types in Turkey, including content blocking, throttling, shutdowns, selective service blocking, filtering, and other reported restriction methods.
- Content-level controls
- Network-level disruptions
- Platform / event-specific restrictions
The data reflects reported incidents and does not represent a comprehensive record of all restrictions.
Restriction types were grouped based on the level at which interference occurs: content-level controls, network-level disruptions, and platform- or event-specific restrictions. Grouping is intended to highlight long-term shifts in the form of internet interference rather than granular differences between individual methods.
Source: Access Now, NetBlocks / TurkeyBlocks, Freedom House, and publicly reported incidents.
Show the data
| Years | Content-level controls | Network-level disruptions | Platform / event-specific restrictions |
|---|---|---|---|
| 2007 | 1 | 0 | 0 |
| 2008 | 1 | 0 | 0 |
| 2009 | 1 | 0 | 0 |
| 2010 | 4 | 0 | 0 |
| 2011 | 3 | 0 | 0 |
| 2012 | 6 | 0 | 1 |
| 2013 | 2 | 2 | 0 |
| 2014 | 3 | 1 | 0 |
| 2015 | 0 | 0 | 0 |
| 2016 | 0 | 11 | 1 |
| 2017 | 0 | 0 | 2 |
| 2018 | 1 | 1 | 2 |
| 2019 | 0 | 2 | 0 |
| 2020 | 0 | 1 | 0 |
| 2021 | 0 | 0 | 0 |
| 2022 | 0 | 1 | 0 |
| 2023 | 1 | 1 | 0 |
| 2024 | 0 | 2 | 0 |
| 2025 | 0 | 2 | 0 |
Given that the data is not comprehensive, we cannot assume when Turkey first started implementing which type of restriction. However, we can say that:
- 2016 stands out as the year with the most recorded incidents,
- and that before 2016, internet controls mainly involved targeted content/URL blocking; after 2016, full shutdowns and throttling became the dominant methods.
When it came to justification, in most cases, very few or no explanations were provided by BTK for the restrictions. Our datasets, however, include both actual and possible causes behind each reported restriction as documented by our sources (AccessNow, NetBlocks, Turkey Blocks). Our analysis of that data revealed “information control” as the most frequent cause for internet restrictions throughout the years.
Documented Internet Restrictions by Type Over Time
This chart presents the number of documented internet shutdowns and access restrictions in Turkey per year, categorized by the reported or stated reason for the restriction, as recorded by monitoring and reporting sources.
- Information control
- Political repression
- Conflict
- Security incident
- Content regulation
- Copyright enforcement
- Regulatory enforcement
The data reflects reported incidents and does not represent a comprehensive record of all restrictions.
Source: Access Now, NetBlocks / TurkeyBlocks, Freedom House, and publicly reported incidents.
Show the data
| Year | Information control | Political repression | Conflict | Security incident | Content regulation | Copyright enforcement | Regulatory enforcement |
|---|---|---|---|---|---|---|---|
| 2007 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2008 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2009 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
| 2010 | 3 | 2 | 0 | 0 | 0 | 0 | 0 |
| 2011 | 0 | 2 | 0 | 0 | 0 | 1 | 0 |
| 2012 | 3 | 1 | 0 | 0 | 3 | 0 | 0 |
| 2013 | 3 | 1 | 0 | 0 | 0 | 0 | 0 |
| 2014 | 3 | 1 | 0 | 0 | 0 | 0 | 0 |
| 2015 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2016 | 4 | 4 | 0 | 4 | 0 | 0 | 0 |
| 2017 | 1 | 0 | 0 | 0 | 0 | 0 | 2 |
| 2018 | 3 | 0 | 0 | 0 | 0 | 1 | 0 |
| 2019 | 0 | 0 | 2 | 0 | 0 | 0 | 0 |
| 2020 | 0 | 0 | 1 | 0 | 0 | 0 | 0 |
| 2021 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2022 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2023 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2024 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2025 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
An example we came across in our data is the Istiklal Street explosion incident in November 2022. BTK was reported to have throttled access to Twitter, Instagram, Facebook, YouTube, and Telegram, reportedly to control the flow of information and manage public perception.
Turkey throttles internet access following deadly explosion in Istanbul Global Voices Advox · 13 November 2022
The second most frequent cause was “political repression.” A notable example was a regional shutdown in September 2016, 2 months post-coup, that affected Turkey’s Southeastern, largely ethnic-Kurdish region, at the same time of removal of 28 elected mayors from office. They were removed using an emergency law that came into force earlier that month after July's failed coup, as reported by BBC. For about 4 hours, residents of that region experienced a full-blown internet and electricity blackout. Turkey Blocks analysts have stated that they believe the shutdown was facilitated by the then-new kill-switch law (see clause 10 above) with the intent to “to suppress voices of dissent or opposition.”
Turkey Blocks on X 26 August 2016
Finding Correlation – The OONI data
Disclaimer: The Open Observatory of Network Interference started conducting connectivity tests for Turkey in 2016. Since we do not have the same network-level, systematic data for earlier years, we cannot merge this dataset with the others to form one conclusion, hence the separation. The OONI data, moreover, does not include reasons for connectivity test failure. But, what we can do is use this data to validate reported incidents and identify patterns in the types of interference.
Increase in Connectivity Failures in Turkey
Since 2016, there is a noticeable increase in connectivity issues in Turkey recorded for certain web and app categories.
Using OONI data aggregated by web category, this chart clearly shows a noticeable increase in the number of failed access attempts to websites across the web in categories from Social Media to News Sites. The question of Why leaves room for speculation.
- Communication Apps
- Social Networks and Media
- Human Rights
- News Platforms
- Political Criticism Blogs
- Average
Share of OONI web connectivity tests in Turkey that recorded an anomaly, by website category. Anonymization tools are left out here so the other categories stay readable; they appear in the additional visualizations below.
Source: OONI
Show the data
| Year | Communication Apps | Social Networks and Media | Human Rights | News Platforms | Political Criticism Blogs | Average |
|---|---|---|---|---|---|---|
| 2016 | 1.02 | – | – | – | – | 1.02 |
| 2017 | 1.68 | 3.37 | 1.80 | 2.18 | 1.59 | 2.12 |
| 2018 | 0.52 | 0.43 | 1.13 | 0.74 | 0.63 | 0.69 |
| 2019 | 1.33 | 7.00 | 1.94 | 2.10 | 4.89 | 3.45 |
| 2020 | 6.53 | 5.12 | 6.56 | 11.50 | 9.30 | 7.80 |
| 2021 | 1.95 | 1.76 | 3.85 | 12.46 | 8.59 | 5.72 |
| 2022 | 1.81 | 2.14 | 3.80 | 9.58 | 8.84 | 5.24 |
| 2023 | 1.33 | 3.24 | 3.79 | 6.61 | 8.91 | 4.78 |
| 2024 | 1.98 | 1.30 | 3.80 | 10.53 | 13.07 | 6.14 |
| 2025 | 3.10 | 3.56 | 6.12 | 8.93 | 10.72 | 6.49 |
Looking at the OONI data, 2016 does not appear to be an isolated incident of “extraordinary measures caused by extraordinary circumstances,” but rather marks a new baseline for internet restrictions in Turkey, a reference point against which later internet restrictions can be compared.
It is also worth noting that while the state of emergency that gave birth to the internet “kill-switch” regulation ended in 2018, the regulatory powers granted to BTK at that time remain in effect.
Taken together, the post-2016 data shows not only continuity but, most importantly, stabilization in the use of internet restrictions. The application of the “kill-switch” law, first introduced under emergency conditions, in subsequent years, appears increasingly detached from the exceptional conditions under which it was originally introduced.
Internet restrictions have come to appear in recognizable sets of political conditions: protests, conflict, security threats, and elections, among others. The recurrence of these restrictions and their similar forms (throttling, shutdowns, platform-specific blocks, etc.) allow us to see this practice as a new pattern, an established administrative practice, rather than isolated responses.
The observed pattern has unfolded alongside political discourse that has, at times, been openly hostile to digital media. For instance, President Recep Tayyip Erdogan had stated in 2014:
I am increasingly against the internet every day,
while vowing to take Twitter down.
Watch the video YouTube
Notably, our analysis identifies Twitter/X as the platform most frequently affected by internet restrictions.
Online Platforms and Services Documented as Affected by Internet Restrictions
This chart lists online platforms, services, and websites that have been documented as affected by internet shutdowns, throttling, or blocking, based on publicly reported incidents from 2007 to 2025.
The data reflects reported incidents and does not represent a comprehensive record of all restrictions.
Source: Access Now, NetBlocks / TurkeyBlocks, Freedom House, and publicly reported incidents.
Show the data
| Platform | Incidents |
|---|---|
| Twitter / X | 18 |
| YouTube | 14 |
| 11 | |
| 10 | |
| 5 | |
| News websites (general) | 4 |
| Telegram | 3 |
| Google services | 3 |
| Dropbox | 2 |
| Wikipedia | 2 |
| Blogspot | 2 |
| Discord | 1 |
| Grindr | 1 |
| SoundCloud | 1 |
| WordPress | 1 |
| Vimeo | 1 |
| 1 | |
| GitHub | 1 |
While such statements and such data do not guarantee direct causality, they do form part of the broader context in which expansive internet control has been exercised.
As of 2026, major platforms remain blocked by court order, and the regulatory authorities introduced in 2016 remain intact. What began as an emergency response has not been formally reversed, allowing internet restrictions to persist as an available and recurrent instrument of governance.
Additional visualizations
Result of the elaborate OONI analysis we did in terms of connection issues for single apps.
App Connection Issues in Turkey
Popular Communication Apps
This chart shows the number of conntection issues the internet control watchdog OONI recorded over the period of 2025 per app. From the highlights and the spikes there is a clear correlation we can draw from politically relevant events and the number of connection issues. Some more apparent than others.
- Twitter/X
- Signal
- Other apps combined
The data shown in this graph is a merge of the app connectivity and web connectivity data provided by OONI for Turkey in 2025.
Other apps combined: Discord, Facebook, Facebook Messenger, Instagram, Telegram, Tinder, WhatsApp, YouTube, Reddit. Each app is listed separately in the data table.
Source: OONI
Show the data
| Month | Discord | Facebook Messenger | Signal | Telegram | Tinder | Twitter/X | YouTube | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Jan | 0 | 3 | 0 | 2 | 1,475 | 14 | 0 | 1,021 | 0 | 2 | 0 |
| Feb | 0 | 62 | 5 | 55 | 1,144 | 7 | 8 | 950 | 5 | 12 | 12 |
| Mar | 0 | 26 | 0 | 15 | 1,188 | 2 | 0 | 1,042 | 10 | 12 | 0 |
| Apr | 4 | 18 | 0 | 17 | 814 | 5 | 0 | 937 | 23 | 6 | 1 |
| May | 30 | 10 | 0 | 30 | 402 | 2 | 0 | 790 | 32 | 9 | 2 |
| Jun | 38 | 151 | 13 | 136 | 76 | 28 | 20 | 782 | 44 | 32 | 44 |
| Jul | 36 | 24 | 4 | 20 | 7 | 6 | 0 | 744 | 7 | 16 | 1 |
| Aug | 69 | 82 | 22 | 86 | 28 | 25 | 2 | 664 | 24 | 62 | 7 |
| Sep | 38 | 120 | 23 | 121 | 27 | 30 | 4 | 811 | 26 | 68 | 5 |
| Oct | 116 | 20 | 0 | 40 | 10 | 4 | 0 | 554 | 1 | 10 | 0 |
| Nov | 32 | 9 | 0 | 10 | 1 | 4 | 0 | 561 | 4 | 10 | 2 |
| Dec | 0 | 22 | 0 | 14 | 9 | 4 | 0 | 282 | 5 | 4 | 0 |
Result of the elaborate OONI analysis about connection issues in only 2025.
App Connection Issues in Turkey
Correlation with Political Events
This chart highlights majoy political events. What can be seen from the charts is a clear correlation between the increase in connection issues for certain apps and the throttling of internet services for "national safety reasons".
- Facebook Messenger
- Discord
- YouTube
- Telegram
The data shown in this graph is a merge of the app connectivity data provided by OONI for Turkey in 2025. In this dataset, Twitter and Signal were excluded since they have been the target of ongoing and repeated blocks by the Turkish government.
Source: OONI
Show the data
| Month | Facebook Messenger | Discord | YouTube | Telegram | |||
|---|---|---|---|---|---|---|---|
| Jan | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
| Feb | 5 | 10 | 7 | 5 | 0 | 6 | 5 |
| Mar | 0 | 9 | 4 | 0 | 0 | 6 | 0 |
| Apr | 0 | 8 | 5 | 0 | 1 | 3 | 0 |
| May | 0 | 4 | 4 | 0 | 15 | 4 | 0 |
| Jun | 13 | 19 | 17 | 13 | 19 | 16 | 13 |
| Jul | 4 | 11 | 6 | 4 | 18 | 8 | 4 |
| Aug | 22 | 30 | 32 | 22 | 34 | 31 | 22 |
| Sep | 23 | 53 | 33 | 23 | 19 | 34 | 22 |
| Oct | 0 | 7 | 6 | 0 | 58 | 5 | 0 |
| Nov | 0 | 4 | 1 | 0 | 16 | 5 | 0 |
| Dec | 0 | 5 | 1 | 1 | 0 | 2 | 0 |
OONI issues by category (same data as in the data story).
Connection Issues by Category
In Percent of Test Requests made
Darker cells mean a higher share of failed tests. A dash means no tests were recorded for that category.
Source: OONI
Show the data
| Year | Anonymization Tools | Communication Apps | Social Networks and Media | Human Rights | News Platforms | Political Criticism Blogs |
|---|---|---|---|---|---|---|
| 2016 | 10 | 1.02 | – | – | – | – |
| 2017 | 12.31 | 1.68 | 3.37 | 1.80 | 2.18 | 1.59 |
| 2018 | 5.85 | 0.52 | 0.43 | 1.13 | 0.74 | 0.63 |
| 2019 | 9.32 | 1.33 | 7.00 | 1.94 | 2.10 | 4.89 |
| 2020 | 20.04 | 6.53 | 5.12 | 6.56 | 11.50 | 9.30 |
| 2021 | 12.68 | 1.95 | 1.76 | 3.85 | 12.46 | 8.59 |
| 2022 | 12.36 | 1.81 | 2.14 | 3.80 | 9.58 | 8.84 |
| 2023 | 14.13 | 1.33 | 3.24 | 3.79 | 6.61 | 8.91 |
| 2024 | 19.62 | 1.98 | 1.30 | 3.80 | 10.53 | 13.07 |
| 2025 | 19.49 | 3.10 | 3.56 | 6.12 | 8.93 | 10.72 |