The State of Internet Censorship in Thailand
Image: Block page in Thailand
A research study by the Open Observatory of Network Interference (OONI), Sinar Project, and the Thai Netizen Network.
Table of contents
Probed ISPs: Triple-T Internet (AS45758), True Internet Co. Ldt (AS17552), JasTel Network International Gateway (AS45629), Realmove Company Limited (AS132061), Advanced Wireless Network Company Limited (AS131445), Symphony Communication (Thailand) PLC. (AS132280), AIS Fibre (AS133481), TOT Public Company Limited (AS23969), Total Access Communication PLC. (AS24378), CAT TELECOM Public Company Ldt. (AS131090), UIH/ BB Broadband (AS38794), TRUE INTERNET Co., Ldt. (AS7470), SBN-ISP/AWN-ISP, maintained by Advanced Wireless Network Company Limited (AS45458), DTAC Broadband (AS132032), The Communication Authority of Thailand, CAT (AS9931), TOT Public Company Limited (AS56120).
OONI tests: Web Connectivity, HTTP Invalid Request Line, HTTP Header Field Manipulation, Vanilla Tor, WhatsApp, Facebook Messenger.
Testing period: 6th November 2016 - 27th February 2017.
Censorship method: DNS hijacking, HTTP transparent proxies (delivering block pages).
New OONI data reveals the blocking of 13 websites in Thailand across 6 different ISPs, between 6th November 2016 and 27th February 2017. Thai ISPs appear to primarily be implementing censorship through DNS hijacking and through the use of middle boxes (HTTP transparent proxies) which serve block pages.
The blocked sites include:
News outlets (nypost.com and dailymail.co.uk)
Anonymity and censorship circumvention tool sites (e.g. hotspotshield.com)
Since these sites were not found to be blocked across all 16 ISPs where tests were run, service providers in Thailand may be in a position to filter online content at their own discretion.
WhatsApp, Facebook Messenger, and the Tor network appear to have been accessible across all tested networks throughout the testing period. Previously blocked sites, such as prachatai.com, were also found to be accessible.
Multiple censorship events in Thailand have been reported over the last decade. More than 10,000 URLs were reportedly blocked in 2010 on the grounds of national security. Further restrictions on freedom of speech and the press appear to have taken place following Thailand’s most recent coup d’etat, as reported by the Citizen Lab which found 56 websites to be blocked between May to June 2014. Independent news outlets, such as Prachatai, have also been blocked in the past.
In an attempt to examine the current state of internet censorship in Thailand, the Open Observatory of Network Interference (OONI), Sinar Project and the Thai Netizen Network collaborated on a joint study to examine whether internet censorship events were persisting in the country through the collection and analysis of network measurements.
The aim of this study is to increase transparency of internet controls in Thailand and to collect data that can potentially corroborate rumours and reports of internet censorship events. The following sections of this report provide information about Thailand’s network landscape and internet penetration levels, its legal environment with respect to freedom of expression, access to information and privacy, as well as about cases of censorship and surveillance that have previously been reported in the country. The remainder of the report documents the methodology and key findings of this study.
Thailand is a constitutional monarchy in Southeast Asia with a population of around 68 million. Geographically, it is bordered to the north by Myanmar and Laos, to the east by Laos and Cambodia, to the south by the Gulf of Thailand and Malaysia, and to the west by the Andaman Sea.
The majority of Thailand’s population is made up by the Thais (95.9%), the Burmese (2%), and the remaining 1.3% constituting of other ethnic groups. Thailand’s official religion is Buddhism with 93.6% of its population identifying as Buddhists. Muslims constitute the second largest religious group (4.9%), followed by Christians (1.2%).
Historically, Thailand has had a strong economy due to its free-enterprise economy, well-developed infrastructure and generally pro-investment policies. However, it has experienced slow growth from 2013 to 2015 as a result of its domestic political turmoil. Thailand ranked low in the 2016 Corruption Perceptions Index, ranking 101 out of 176. Human rights violations and systematic denial of basic rights such as freedom of expression, association, and assembly, occurred with regularity during the military junta’s rule of the country since 2014.
Politically, Thailand is a constitutional monarchy where the Prime Minister is the head of government and a hereditary monarch is the head of state. However, Thailand has oscillated between being ruled by a parliamentary democracy and military junta for decades, with the latest military coup being in May 2014 in which the army chief, General Prayut Chan-o-cha, was elected as Prime Minister and the National Council for Peace and Order (NCPO) was established.
Unlike previous coups, there have been significant delays in the return to civilian rule. Promises had been made by the NCPO to hold new elections only after a new constitution was enacted. However, the general elections have repeatedly been delayed by events such as the rejection of the initial draft by government officials in 2015, issues of succession to the throne of Thailand following the Thai monarch’s death at the end of 2016, and the newly crowned King Vajiralongkorn’s refusal to put the new constitution into effect until amendments were made. Such proposed provisions would be aimed at expanding his powers, from allowing him to spend time abroad without the appointment of a regent, to absolving the need for a countersignature on all royal acts which would give him the disproportionate power of signing executive orders and decrees individually. Currently, general elections in Thailand are expected to be held in mid-2018.
Network landscape and internet penetration
The Thai government has long held expansive control over the internet, largely facilitated by its various relationships with Internet Service Providers (ISPs) and telecommunication companies. Amongst more than ten International Internet Gateways (IIGs) in the country, CAT and TOT,both state-owned, are still two of the largest. Former politicians, military officers, or members of their families also hold key positions in large telecommunication companies. Under telecommunication law, all service providers are subjected to license suspensions or revocation if found not cooperating with the regulator in law enforcement, which also includes network shutdowns. More frequently, ISPs are asked “informally” to block certain content.
The internet and mobile service providers of Thailand are a mix of state-owned companies and private operators. The three fixed line operators in Thailand are True Corporation, TT&T, and the state-owned TOT. The number of fixed lines is gradually declining with the expansion of mobile phone services in Thailand which have experienced significant growth with market penetration peaking at 146% in 2014, and declining to 128% in 2016 due to market consolidation. The three major private mobile carriers are AIS, DTAC, and TrueMove.
|Mobile Operators/ISPs||Fixed Internet||Mobile Internet|
|Telephone Organization of Thailand Public Company Limited (TOT)||X||X|
|Advanced Info Service Public Company Limited (AIS)||X||X|
|TT&T Public Company Limited||X||X|
As of 2016, 60.1% of Thailand’s population has access to the internet. According to the 2016 Thai Information and Communication Technology Survey in Household, there were 32.3% computer users, 47.5% internet users and 81.4% mobile phone users from a population of 62.8 million aged 6 years and up in 2016. Most Thai internet and smartphone users reside in municipal areas, which have a higher average household income. The proportion in internet use has increased from 37.7% in 2012 to 57.4% in 2016 for municipal areas, and from 20.5% from 2012 to 39.5% in 2016 for non-municipal areas.
Social media is widely utilised in Thailand, with Bangkok topping the global Facebook users list by city, and Siam Paragon, a shopping mall in Bangkok being the most Instagrammed location on Earth. According to the 2016 Thai Information and Communication Technology Survey in Household, 91.5% of internet users utilised the internet for social networking.
Percentages of households with ICT devices from 2012 to 2016 are illustrated below.
|Year||Households||Fixed Telephone (%)||Computer (%)||Internet (%)|
Of the households with internet access in 2016, 70.6% used mobile internet, and 23.3% utilised fixed broadband.
Freedom of expression
Thailand Penal Code, Section 112 (Lèse-majesté)
Section 112 of the Thai Penal Code penalises anyone who “defames, insults or threatens the King, the Queen, the Heir-apparent or the Regent” with a punishment of imprisonment of three to fifteen years. Lèse-majesté defendants are routinely denied bail, and convictions often result in heavy sentences in most cases. Lèse-majesté or defamation complaints can be lodged by any one citizen against another, and such compliments always require formal investigation from authorities. This brings about the potential for abuse in imposing systematic restrictions of information control to limit social mobilisation around key political events.
On 2nd December 2016, BBC Thai published a profile of Thailand’s new King which was shared widely on social media. Some Thais criticized its content for being insulting to the new king, and the Thai Ministry of Digital Economy reportedly blocked a link to the profile on the BBC’s Thai website on the grounds of displaying “inappropriate content”. Two months later, OONI tests found this site to be accessible in tested networks, but it remains unclear how long this site may have been blocked for. In December 2016, the police and some soldiers subsequently visited the BBC’s office in Bangkok, and a Thai democracy activist was temporarily arrested for sharing a link to the BBC profile.
Computer Crime Act
Article 14(1) of the 2007 Computer Crime Act (CCA) penalises individuals found to have uploaded content deemed to be “forged”, “false”, or which is likely to “cause damage to a third party” with an imprisonment of up to 5 years along with a maximum fine of 100,000 Thai Baht. Such broad and ambiguous language opens up the law to abuse. Article 14(1) of the CCA has long been utilised against journalists, activists and internet users for content considered to be damaging by government sanctioned authorities.
The 2017 amendment to article 14(2) of the CCA broadens the scope, incriminating those found guilty of uploading information that would “damage the maintenance of national security, public safety, national economic security or public infrastructure serving national’s public interest or cause panic in the public”, and expanding the power authorities have in abusing a person’s exercise of their protected right to freedom of expression.
Internal Security Act 2008
Under the 2008 Internal Security Act, the ISA would establish The Internal Security Operations Command (ISOC), a situation monitoring center in every province which would have the authority to respond to alleged threats to national security. The ISOC would be authorised to exercise its powers with respect to situations affecting the national security where a state of emergency has not been declared yet, bypassing the role of parliaments and courts in reviewing or approving the necessity of such abuses of power.
Press freedom in Thailand has been severely restricted post military coup following the military junta’s creation of the National Council for Peace and Order (NCPO) tasked with enforcing widespread censorship.
NCPO Order 97⁄2014
The NCPO announced order 97⁄2014, “Cooperating with the Work of the National Council for Peace and Order(NCPO) and the Distribution of News to the Public” which prohibits publication or broadcast of criticisms of the military authorities from print media, radio, TV, and online media. The NCPO has sole discretion in determining what content falls within prohibited categories. Violations of provisions in this announcement could result in prosecution under the law, and the immediate suspension of the publication or program.
Article 5 of the Head NCPO Order 3⁄2015
Under Article 5 of the Head NCPO Order 3⁄2015, NCPO officers are authorised to issue orders prohibiting the distribution of press items, or the sale of any publication or material that is deemed to have the potential to cause public alarm, or which contains false information likely to cause public misunderstanding, or which threatens public order or national security.
Access to information
Official Information Act 1997
The 1997 Official Information Act gives Thai citizens the right to request for the disclosure of official government information from state agencies. However, authorities have the right to reject requests for the disclosure of information under Article 15 of the act, under unclear and overly broad reasons such as the “decline in efficiency of law enforcement, risk to national security, and endangerment of life or safety to any person”. Although citizens have the right to appeal the rejection, appeals submitted to the committee take a long time to be considered and process, making access difficult for citizens.
In addition, the Act does not cover information in the possession of private entities, which has led to ongoing disputes over whether independent public agencies such as the National Anti-Corruption Commission (NACC), the Office of the Election Commission, and the Office of the Auditor General fall under the scope of the Act.
Constitution of the Kingdom of Thailand
The 2007 Constitution of the Kingdom of Thailand provides citizens with the right to privacy. Under Article 35 of the constitution, “A person’s family rights, dignity, reputation or the right of privacy shall be protected. The assertion or circulation of a statement or picture in any manner whatsoever to the public, which violates or affects a person’s family rights, dignity, reputation or the right of privacy, shall not be made except for the case which is beneficial to the public”.
Following the 2014 military coup, an interim constitution was enacted, in which almost all of the provisions of the original constitution were suspended. There is no longer an explicit provision related to the right to privacy.
While there is no comprehensive general data protection law in Thailand, personal data in the public sector is at some level protected by the Official Information Act B.E. 2540 (1997) which obligates state agencies to allow individuals to correct personal data that is maintained by the agencies. Personal information in the private sector, such as credit information, patient data, and telecommunication data, is regulated by sectoral laws, like the 2008 Credit Information Business Act, the 2007 National Health Act, and the 2006 Notification of the National Telecommunications Commission (on measures to protect the rights of telecommunication consumers in the areas of personal data, right to privacy, and freedom of communication via telecommunication networks).
Censorship and surveillance
2017 Computer Crime Act, Article 20
Under Article 20 of the 2017 Computer-related Crime Act (CCA), the “Computer Data Screening Committee”, a 9 member panel appointed by the government would have the power to suggest the court to suppress or remove computer data that is “deemed to be a breach to the public order or moral high ground of the people.” Due to the broadness of this definition, this allows authorities to act as moral crusaders, giving them a wide latitude to suppress online content that does not violate any laws, but that they deem to be a breach of public morals.
2017 Computer Crime Act, Article 18
Articles 18(2) and 18(3) of the 2017 Computer Crime Act (CCA) would allow user-related data and traffic data to be accessed by authorities without a court order under probable cause to assist with investigations related to an offense under the CCA or other laws.
Article 18(7) would allow authorities with a court order to compel service providers in assisting with the decryption of encoded data, undermining the use of encryption tools as a protection of user privacy.
Reported cases of internet censorship and surveillance
Multiple cases of internet censorship and surveillance have been reported in Thailand over the last decade. According to the Thai Netizen Network, more than 10,000 URLs were blocked in 2010 on the grounds of national security, even though many of them expressed criticism towards the Abhisit Vejjajiva administration. Prachatai, an independent news outlet, was amongst the many sites that were blocked without transparency.
Below we highlight a few cases of internet censorship and surveillance in Thailand, as reported over the last few years.
Censorship following Thailand’s 2014 military coup
Thailand’s most recent May 2014 coup d’etat signaled further restrictions on freedom of speech, association, and the press. According to network measurement tests performed by the Citizen Lab between May to June 2014, 56 URLs were found to be blocked.These sites included domestic independent news outlets, international media critical of the coup, social media accounts sharing anti- coup material, as well as censorship circumvention tools. Facebook was blocked for approximately 40 minutes on 28th May 2014, possibly in an attempt to stop the spread of anti-coup messages. A report by Privacy International however suggests that Facebook may have accidentally been blocked in an attempt to circumvent SSL encryption which would direct traffic over unencrypted HTTP instead of HTTPS, enabling government spying efforts.
Since the 2014 military coup, access to political and social content has increasingly being blocked in Thailand on the grounds of national security and lèse majesté, according to Freedom House. The Royal Thai Police, the Communications Authority of Thailand, and the Ministry of Information and Communication Technology (MICT) have regularly arrested activists and internet users under lèse majesté (Section 112 of the Thai Penal Code) for criticisms of the monarchy, and Article 14 of the Computer Crime Act (CCA) which relates to content deemed to affect “national security”, in addition to imposing systematic restrictions of information control to limit social mobilisation around key political events.
Hacking Team Surveillance Software
Leaked Hacking Team emails in 2015 revealed that the Royal Thai Army and the Corrections Department of the Royal Thai Police had spent €286,482 and €360,000 respectively in 2014 to purchase a surveillance program called Remote Control System (RCS) from Hacking Team, an Italian spyware company. The spyware is designed to monitor the communications of internet users, evade encryption and remotely collect information from a target’s computer. In their correspondence, the National Security Council had specifically asked Hacking Team if their product was capable of targeting LINE, WeChat, and WhatsApp (instant messaging apps used widely in Thailand).
Microsoft’s assistance in Thai government surveillance by omission
According to a Privacy International report, the Thai government has the potential to misuse their root certificate and impersonate an intended website with a falsified certificate to intercept apparently secure communications or for the injection of false, malicious content such as malware. Microsoft was the only certificate authority entitled to sign a root certificate that included the Thai national root certificate in their OS and browser by default.
A spokesperson from Microsoft defended the decision, citing their “extensive review process that includes regular audits from a third-party web trust auditor”. Microsoft’s decision would leave Thai Windows users vulnerable to government surveillance should the Thai government choose to misuse their root certificate.
Technology Crime Suppression Division (TCSD)
Aside from the use of surveillance technology, Thailand’s Information and Communication Technology Ministry has dedicated human resources invested in the monitoring of online activities, with a longstanding 30-person Technology Crime Suppression Division (TCSD) team which scans online posts and follows up on public complaints on cyber crimes and lèse-majesté content, according to the former Minister of ICT in August 2015.
Cyber Scout program
In another instance of state-sponsored surveillance, the Cyber Scout program, initially launched in 2010 by the Ministry of Justice and Ministry of ICT was aimed at recruiting and training students to monitor and report online content that could be deemed as offensive, or a threat to national security, particularly lèse-majesté content. The Cyber Scout program was reintroduced following the 2014 military coup with the Ministry of ICT cooperation with 200 schools, training school children to restrict criticism and dissent by the military junta’s values. By 2015, the program had recruited over 120,000 cyber scouts nationwide, spanning across 88 schools.
Examining internet censorship in Thailand
The Open Observatory of Network Interference (OONI), in collaboration with Sinar Project and the Thai Netizen Network, performed a study of internet censorship in Thailand. The aim of this study was to understand whether and to what extent censorship events occurred in Thailand during the testing period.
The sections below document the methodology and key findings of this study.
The methodology of this study, in an attempt to identify potential internet censorship events in Thailand, included the following:
Review of the Citizen Lab’s Thai test list
OONI network measurements
A list of URLs that are relevant and commonly accessed in Thailand was created by the Citizen Lab in 2014 for the purpose of enabling network measurement researchers to examine their accessibility in Thailand. As part of this study, this list of URLs was reviewed to include additional URLs which - along with other URLs that are commonly accessed around the world - were tested for blocking based on OONI’s free software tests. Such tests were run from local vantage points in Thailand, and they also examined whether systems that are responsible for censorship, surveillance and traffic manipulation were present in the tested network. Once network measurement data was collected from these tests, the data was subsequently processed and analyzed based on a set of heuristics for detecting internet censorship and traffic manipulation.
The testing period for this study started on 6th November 2016 and concluded on 27th February 2017.
Review of the Citizen Lab’s Thai test list
An important part of identifying censorship is determining which websites to examine for blocking.
OONI’s software (called ooniprobe) is designed to examine URLs contained in specific lists (“test lists”) for censorship. By default, ooniprobe examines the “global test list”, which includes a wide range of internationally relevant websites, most of which are in English. These websites fall under 31 categories, ranging from news media, file sharing and culture, to provocative or objectionable categories, like pornography, political criticism, and hate speech.
These categories help ensure that a wide range of different types of websites are tested, and they enable the examination of the impact of censorship events (for example, if the majority of the websites found to be blocked in a country fall under the “human rights” category, that may have a bigger impact than other types of websites being blocked elsewhere). The main reason why objectionable categories (such as “pornography” and “hate speech”) are included for testing is because they are more likely to be blocked due to their nature, enabling the development of heuristics for detecting censorship elsewhere within a country.
In addition to testing the URLs included in the global test list, ooniprobe is also designed to examine a test list which is specifically created for the country that the user is running ooniprobe from, if such a list exists. Unlike the global test list, country-specific test lists include websites that are relevant and commonly accessed within specific countries, and such websites are often in local languages. Similarly to the global test list, country-specific test lists include websites that fall under the same set of 31 categories, as explained previously.
All test lists are hosted by the Citizen Lab on GitHub, supporting OONI and other network measurement projects in the creation and maintenance of lists of URLs to test for censorship. As part of this study, OONI reviewed the Citizen Lab’s test list for Thailand by adding more URLs to be tested for censorship. Overall, 420 URLs that are relevant to Thailand were tested as part of this study. In addition, the URLs included in the Citizen Lab’s global list (including 1,105 different URLs) were also tested.
It is important to acknowledge that the findings of this study are only limited to the websites that were tested, and do not necessarily provide a complete view of other censorship events that may have occurred during the testing period.
OONI network measurements
The Open Observatory of Network Interference (OONI) is a free software project that aims to increase transparency of internet censorship around the world. Since 2012, OONI has developed multiple free and open source software tests designed to examine the following:
Blocking of websites.
Blocking of censorship circumvention tools (such as Tor).
Blocking of instant messaging apps.
Detection of systems responsible for censorship, surveillance, and traffic manipulation.
As part of this study, the following OONI software tests were run from 16 different local vantage points in Thailand:
The Web Connectivity test was run with the aim of examining whether a set of URLs (included in both the “global test list” and the recently updated “Thai test list”) were blocked during the testing period and if so, how. The Vanilla Tor test was run to examine the reachability of the Tor network, while the WhatsApp and Facebook Messenger tests were run to examine whether these instant messaging apps were blocked in Thailand during the testing period.
The HTTP invalid request line and HTTP header field manipulation tests were run with the aim of examining whether “middle boxes” (systems placed in the network between the user and a control server) that could potentially be responsible for censorship and/or surveillance were present in the tested networks.
The sections below document how each of these tests are designed for the purpose of detecting cases of internet censorship and traffic manipulation.
Web Connectivity test
This test examines whether websites are reachable and if they are not, it attempts to determine whether access to them is blocked through DNS tampering, TCP/IP blocking or by a transparent HTTP proxy. Specifically, this test is designed to perform the following:
HTTP GET request
By default, this test performs the above (excluding the first step, which is performed only over the network of the user) both over a control server and over the network of the user. If the results from both networks match, then there is no clear sign of network interference; but if the results are different, the websites that the user is testing are likely censored.
Further information is provided below, explaining how each step performed under the web connectivity test works.
1. Resolver identification
The domain name system (DNS) is what is responsible for transforming a host name (e.g. torproject.org) into an IP address (e.g. 126.96.36.199). Internet Service Providers (ISPs), amongst others, run DNS resolvers which map IP addresses to hostnames. In some circumstances though, ISPs map the requested host names to the wrong IP addresses, which is a form of tampering.
As a first step, the web connectivity test attempts to identify which DNS resolver is being used by the user. It does so by performing a DNS query to special domains (such as whoami.akamai.com) which will disclose the IP address of the resolver.
2. DNS lookup
Once the web connectivity test has identified the DNS resolver of the user, it then attempts to identify which addresses are mapped to the tested host names by the resolver. It does so by performing a DNS lookup, which asks the resolver to disclose which IP addresses are mapped to the tested host names, as well as which other host names are linked to the tested host names under DNS queries.
3. TCP connect
The web connectivity test will then try to connect to the tested websites by attempting to establish a TCP session on port 80 (or port 443 for URLs that begin with HTTPS) for the list of IP addresses that were identified in the previous step (DNS lookup).
4. HTTP GET request
As the web connectivity test connects to tested websites (through the previous step), it sends requests through the HTTP protocol to the servers which are hosting those websites. A server normally responds to an HTTP GET request with the content of the webpage that is requested.
Comparison of results: Identifying censorship
Once the above steps of the web connectivity test are performed both over a control server and over the network of the user, the collected results are then compared with the aim of identifying whether and how tested websites are tampered with. If the compared results do not match, then there is a sign of network interference.
Below are the conditions under which the following types of blocking are identified:
DNS blocking: If the DNS responses (such as the IP addresses mapped to host names) do not match.
TCP/IP blocking: If a TCP session to connect to websites was not established over the network of the user.
HTTP blocking: If the HTTP request over the user’s network failed, or the HTTP status codes don’t match, or all of the following apply:
The body length of compared websites (over the control server and the network of the user) differs by some percentage
The HTTP headers names do not match
The HTML title tags do not match
It’s important to note, however, that DNS resolvers, such as Google or a local ISP, often provide users with IP addresses that are closest to them geographically. Often this is not done with the intent of network tampering, but merely for the purpose of providing users with localized content or faster access to websites. As a result, some false positives might arise in OONI measurements. Other false positives might occur when tested websites serve different content depending on the country that the user is connecting from, or in the cases when websites return failures even though they are not tampered with.
HTTP Invalid Request Line test
This test tries to detect the presence of network components (“middle box”) which could be responsible for censorship and/or traffic manipulation.
Instead of sending a normal HTTP request, this test sends an invalid HTTP request line - containing an invalid HTTP version number, an invalid field count and a huge request method – to an echo service listening on the standard HTTP port. An echo service is a very useful debugging and measurement tool, which simply sends back to the originating source any data it receives. If a middle box is not present in the network between the user and an echo service, then the echo service will send the invalid HTTP request line back to the user, exactly as it received it. In such cases, there is no visible traffic manipulation in the tested network.
If, however, a middle box is present in the tested network, the invalid HTTP request line will be intercepted by the middle box and this may trigger an error and that will subsequently be sent back to OONI’s server. Such errors indicate that software for traffic manipulation is likely placed in the tested network, though it’s not always clear what that software is. In some cases though, censorship and/or surveillance vendors can be identified through the error messages in the received HTTP response. Based on this technique, OONI has previously detected the use of BlueCoat, Squid and Privoxy proxy technologies in networks across multiple countries around the world.
It’s important though to note that a false negative could potentially occur in the hypothetical instance that ISPs are using highly sophisticated censorship and/or surveillance software that is specifically designed to not trigger errors when receiving invalid HTTP request lines like the ones of this test. Furthermore, the presence of a middle box is not necessarily indicative of traffic manipulation, as they are often used in networks for caching purposes.
HTTP Header Field Manipulation test
This test also tries to detect the presence of network components (“middle box”) which could be responsible for censorship and/or traffic manipulation.
HTTP is a protocol which transfers or exchanges data across the internet. It does so by handling a client’s request to connect to a server, and a server’s response to a client’s request. Every time a user connects to a server, the user (client) sends a request through the HTTP protocol to that server. Such requests include “HTTP headers”, which transmit various types of information, including the user’s device operating system and the type of browser that is being used. If Firefox is used on Windows, for example, the “user agent header” in the HTTP request will tell the server that a Firefox browser is being used on a Windows operating system.
This test emulates an HTTP request towards a server, but sends HTTP headers that have variations in capitalization. In other words, this test sends HTTP requests which include valid, but non-canonical HTTP headers. Such requests are sent to a backend control server which sends back any data it receives. If OONI receives the HTTP headers exactly as they were sent, then there is no visible presence of a “middle box” in the network that could be responsible for censorship, surveillance and/or traffic manipulation. If, however, such software is present in the tested network, it will likely normalize the invalid headers that are sent or add extra headers.
Depending on whether the HTTP headers that are sent and received from a backend control server are the same or not, OONI is able to evaluate whether software – which could be responsible for traffic manipulation – is present in the tested network.
False negatives, however, could potentially occur in the hypothetical instance that ISPs are using highly sophisticated software that is specifically designed to not interfere with HTTP headers when it receives them. Furthermore, the presence of a middle box is not necessarily indicative of traffic manipulation, as they are often used in networks for caching purposes.
Vanilla Tor test
The Vanilla Tor test attempts to start a connection to the Tor network. If the test successfully bootstraps a connection within a predefined amount of seconds (300 by default), then Tor is considered to be reachable from the vantage point of the user. But if the test does not manage to establish a connection, then the Tor network is likely blocked within the tested network.
This test is designed to examine the reachability of both WhatsApp’s app and the WhatsApp web version within a network.
OONI’s WhatsApp test attempts to perform an HTTP GET request, TCP connection and DNS lookup to WhatsApp’s endpoints, registration service and web version over the vantage point of the user. Based on this methodology, WhatsApp’s app is likely blocked if any of the following apply:
TCP connections to WhatsApp’s endpoints fail;
TCP connections to WhatsApp’s registration service fail;
DNS lookups resolve to IP addresses that are not allocated to WhatsApp;
HTTP requests to WhatsApp’s registration service do not send back a response to OONI’s servers.
WhatsApp’s web interface (web.whatsapp.com) is likely if any of the following apply:
TCP connections to web.whatsapp.com fail;
DNS lookups illustrate that a different IP address has been allocated to web.whatsapp.com;
HTTP requests to web.whatsapp.com do not send back a consistent response to OONI’s servers.
Facebook Messenger test
This test is designed to examine the reachability of Facebook Messenger within a tested network.
OONI’s Facebook Messenger test attempts to perform a TCP connection and DNS lookup to Facebook’s endpoints over the vantage point of the user. Based on this methodology, Facebook Messenger is likely blocked if one or both of the following apply:
TCP connections to Facebook’s endpoints fail;
DNS lookups to domains associated to Facebook do not resolve to IP addresses allocated to Facebook.
Through its data pipeline, OONI processes all network measurements that it collects, including the following types of data:
OONI by default collects the code which corresponds to the country from which the user is running ooniprobe tests from, by automatically searching for it based on the user’s IP address through the MaxMind GeoIP database. The collection of country codes is an important part of OONI’s research, as it enables OONI to map out global network measurements and to identify where network interferences take place.
Autonomous System Number (ASN)
OONI by default collects the Autonomous System Number (ASN) which corresponds to the network that a user is running ooniprobe tests from. The collection of the ASN is useful to OONI’s research because it reveals the specific network provider (such as Vodafone) of a user. Such information can increase transparency in regards to which network providers are implementing censorship or other forms of network interference.
Date and time of measurements
OONI by default collects the time and date of when tests were run. This information helps OONI evaluate when network interferences occur and to compare them across time.
IP addresses and other information
OONI does not deliberately collect or store users’ IP addresses. In fact, OONI takes measures to remove users’ IP addresses from the collected measurements, to protect its users from potential risks.
However, OONI might unintentionally collect users’ IP addresses and other potentially personally-identifiable information, if such information is included in the HTTP headers or other metadata of measurements. This, for example, can occur if the tested websites include tracking technologies or custom content based on a user’s network location.
The types of network measurements that OONI collects depend on the types of tests that are run. Specifications about each OONI test can be viewed through its git repository, and details about what collected network measurements entail can be viewed through OONI Explorer or through OONI’s measurement API.
OONI processes the above types of data with the aim of deriving meaning from the collected measurements and, specifically, in an attempt to answer the following types of questions:
Which types of OONI tests were run?
In which countries were those tests run?
In which networks were those tests run?
When were tests run?
What types of network interference occurred?
In which countries did network interference occur?
In which networks did network interference occur?
When did network interference occur?
How did network interference occur?
To answer such questions, OONI’s pipeline is designed to process data which is automatically sent to OONI’s measurement collector by default. The initial processing of network measurements enables the following:
Attributing measurements to a specific country.
Attributing measurements to a specific network within a country.
Distinguishing measurements based on the specific tests that were run for their collection.
Distinguishing between “normal” and “anomalous” measurements (the latter indicating that a form of network tampering is likely present).
Identifying the type of network interference based on a set of heuristics for DNS tampering, TCP/IP blocking, and HTTP blocking.
Identifying block pages based on a set of heuristics for HTTP blocking.
Identifying the presence of “middle boxes” within tested networks.
However, false positives can emerge within the processed data due to a number of reasons. As explained previously (section on “OONI network measurements”), DNS resolvers (operated by Google or a local ISP) often provide users with IP addresses that are closest to them geographically. While this may appear to be a case of DNS tampering, it is actually done with the intention of providing users with faster access to websites. Similarly, false positives may emerge when tested websites serve different content depending on the country that the user is connecting from, or in the cases when websites return failures even though they are not tampered with.
Furthermore, measurements indicating HTTP or TCP/IP blocking might actually be due to temporary HTTP or TCP/IP failures, and may not conclusively be a sign of network interference. It is therefore important to test the same sets of websites across time and to cross-correlate data, prior to reaching a conclusion on whether websites are in fact being blocked.
Since block pages differ from country to country and sometimes even from network to network, it is quite challenging to accurately identify them. OONI uses a series of heuristics to try to guess if the page in question differs from the expected control, but these heuristics can often result in false positives. For this reason OONI only says that there is a confirmed instance of blocking when a block page is detected.
OONI’s methodology for detecting the presence of “middle boxes” - systems that could be responsible for censorship, surveillance and traffic manipulation - can also present false negatives, if ISPs are using highly sophisticated software that is specifically designed to not interfere with HTTP headers when it receives them, or to not trigger error messages when receiving invalid HTTP request lines. It remains unclear though if such software is being used. Moreover, it’s important to note that the presence of a middle box is not necessarily indicative of censorship or traffic manipulation, as such systems are often used in networks for caching purposes.
Upon collection of more network measurements, OONI continues to develop its data analysis heuristics, based on which it attempts to accurately identify censorship events.
As part of this study, network measurements were collected through ooniprobe software tests performed across 16 different local vantage points in Thailand between 6th November 2016 to 27th February 2017.
Upon analysis of the collected data, the findings illustrate that ISPs in Thailand are primarily implementing censorship through DNS hijacking and through the use of middle boxes (HTTP transparent proxies) which serve block pages. OONI’s HTTP invalid request line test, in particular, revealed the presence of middle boxes in many networks, which intercepted the HTTP requests that were sent to echo servers. OONI’s Web Connectivity test, on the other hand, revealed that many ISPs served block pages for 13 different sites.
The types of sites that were found to be blocked as part of this study include:
Anonymity and censorship circumvention tools
The table below illustrates all of the sites that we confirmed to be blocked across ISPs as part of our testing and data analysis.
|Internet Service Providers (ISPs)||Blocked websites||Categories||Date of blocking|
|DTAC (AS24378)||http://www.nypost.com||News media||2/23/2017|
|TOT 3BB (AS23969)||http://www.dailymail.co.uk||News media||2/20/2017|
|TOT 3BB (AS23969)||http://www.hotspotshield.com||Anonymity and censorship circumvention||2/20/2017|
|Realmove Company Limited (AS132061)||http://www.wikileaks.org||News media||2/14/2017|
|Realmove Company Limited (AS132061)||http://anonymouse.org||Anonymity and censorship circumvention||2/14/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://ultrasurf.us||Anonymity and censorship circumvention||2/22/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://pridetube.com||Pornography||2/14/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://www.naughty.com||Pornography||2/22/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://www.livejasmin.com||Pornography||2/24/2017|
|True Internet CO., LDT (AS17552)||http://redtube.com||Pornography||2/22/2017|
|True Internet CO., LDT (AS17552)||http://xhamster.com||Pornography||2/16/2017|
|JasTel Network International Gateway (AS45629)||http://youjizz.com||Pornography||2/14/2017|
Following a meeting in early 2015 between Thailand’s National Broadcasting and Telecommunications Commission (NBTC), various Internet Service Providers (ISPs), and the police’s Special Branch, Thai ISPs were “authorised” to block online content at their own discretion. This appears to be corroborated by our findings, which show different sites being blocked by different ISPs across time, indicating that service providers may have flexibility in terms of what they can filter.
New York Post, for example, was only found to be blocked in one mobile network (DTAC), while being accessible across all other ISPs where tests were run. Similarly, WikiLeaks was only found to be blocked by one provider (Realmove Company Limited). This indicates that ISPs were probably not ordered to block WikiLeaks on the grounds of “national security”, but rather that one provider likely chose to block the site at its own discretion.
Anonymouse.org was found to be blocked twice, while the rest of the sites in the table above were found to be blocked only once. However, as the measurements were run quite sporadically across different networks, it remains quite unclear whether and to what extent some of these sites remain blocked.
OONI data shows that DTAC, Thailand’s second largest GSM mobile phone provider, blocked access to nypost.com in February 2017. The site though was found to be accessible across other ISPs. While the company’s motivation remains unclear, it may have chosen to block access to this news outlet if it was publishing information that was viewed as offensive under lese majeste laws. Similarly, TOT 3BB blocked access to dailymail.co.uk, but rather than serving a block page, dailymail.co.uk was found to be blocked based on DNS censorship.
|Internet Service Providers (ISPs)||Blocked media sites||Date of blocking|
|TOT 3BB (AS23969)||http://www.dailymail.co.uk||2/20/2017|
|Realmove Company Limited (AS132061)||http://www.wikileaks.org||2/14/2017|
WikiLeaks is a multi-national media organization that is known for publishing large datasets of restricted official materials involving war, spying, and corruption. In 2008, WikiLeaks released a list of blacklisted websites by Thailand’s Ministry of Information and Communication Technology (MICT). Wikileaks.org was subsequently blocked in Thailand less than two years later under the 2005 emergency decree.
Our findings show that while wikileaks.org was accessible across 15 ISPs during our testing, it was recently found to be blocked in February 2017 by Realmove Company Limited.
Anonymity and censorship circumvention
Other sites, however, for anonymity and censorship circumvention were found to be blocked, as illustrated in the table below.
|Internet Service Providers (ISPs)||Blocked anonymity & circumvention sites||Date of blocking|
|TOT 3BB (AS23969)||http://www.hotspotshield.com||2/20/2017|
|Realmove Company Limited (AS132061)||http://anonymouse.org||2/14/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://ultrasurf.us||2/22/2017|
HotSpot Shield is a free Virtual Private Network (VPN) that enables its users to enhance their online privacy and to circumvent online censorship. The testing of hotspotshield.com showed that TOT 3BB served a block page in February 2017. Similarly, the testing of anonymouse.org and ultrasurf.us also showed that providers (Realmove Company Limited and Triple-T Internet Co., Ldt) served block pages.
Thailand’s Prevention and Suppression of Temptations to Dangerous Behaviors Bill aims to prohibit specific types of pornography. As part of our testing, we found the following pornographic sites to be blocked in Thailand.
|Internet Service Providers (ISPs)||Blocked porn sites||Date of blocking|
|Triple-T Internet Co., Ldt (AS45758)||http://pridetube.com||2/14/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://www.naughty.com||2/22/2017|
|Triple-T Internet Co., Ldt (AS45758)||http://www.livejasmin.com||2/24/2017|
|True Internet CO., LDT (AS17552)||http://redtube.com||2/22/2017|
|True Internet CO., LDT (AS17552)||http://xhamster.com||2/16/2017|
|JasTel Network International Gateway (AS45629)||http://youjizz.com||2/14/2017|
Acknowledgement of limitations
The findings of this study present various limitations and do not necessarily reflect a comprehensive view of internet censorship in Thailand.
The first limitation is associated with the testing period. While OONI network measurements have been collected from Thailand since 2014 and continue to be collected on the day of the publication of this report, this study only takes into account and analyzes network measurements that were collected between 6th November 2016 and 27th February 2017. This study is limited to this time frame because we aim to examine the most recent censorship events and because there was a significant increase in the collection of network measurements during this period, in comparison to previous months and years. As such, censorship events which may have occurred before and/or after the testing period are not examined as part of this study.
Another limitation to this study is associated to the amount and types of URLs that were tested for censorship. As mentioned in the methodology section of this report (“Creating a Thai test list”), OONI’s Web Connectivity test was run to examine the accessibility of 420 URLs that are more relevant to the Thai context and of 1,105 internationally relevant sites. While a total of 1,525 URLs were tested for censorship as part of this study, we did not test all of the URLs on the internet, indicating the possibility that other websites not included in tests lists might have been blocked.
Finally, while network measurements were collected from 16 different local vantage points in Thailand, OONI’s software tests were not run consistently across all networks. Stable measurements were collected from certain vantage points throughout the testing period, but less stable measurements were also collected from a number of other vantage points following the launch of OONI’s mobile app on 9th February 2017. In other words, once ooniprobe became easier to install and run via its mobile version for Android and iOS, we received an increased amount of sporadic measurements from various new networks. In some of these networks we were able to identify additional cases of censorship, but since tests were not always run consistently, our ability to evaluate whether censorship cases were persistent was limited.
Multiple censorship events have been reported in Thailand over the last decade, particularly since the latest military coup in the country in May 2014, involving the blocking of news outlets and sites that express political criticism.
The objective of this study is to gain a better understanding of internet censorship events in Thailand through the collection and analysis of network measurements. To this end, OONI software tests were run across 16 different local vantage points in Thailand with the aim of collecting and analyzing network measurement data that could help examine whether sites, instant messaging apps, and censorship circumvention tools were blocked. Some of the tests that were also run are designed to examine whether systems (“middle boxes”) that could be responsible for censorship, surveillance, and traffic manipulation were present in the tested networks. Overall, the accessibility of 1,525 sites was tested, and the network measurement data collected between 6th November 2016 to 27th February 2017 was analyzed.
The key findings of this study show that Thai ISPs appear to primarily be implementing censorship through DNS hijacking and through the use of middle boxes (HTTP transparent proxies) which serve block pages, while in fewer cases, ISPs appear to be implementing DNS-based censorship (in the case of the blocking of dailymail.co.uk, for example). It’s worth noting that Thai ISPs appear to be implementing censorship at their own discretion, since the types of sites blocked vary across ISPs.
As part of this study, 13 sites were confirmed to be blocked across 6 ISPs (DTAC, Realmove Company Limited, TOT 3BB, Triple-T Internet Co., Ltd, True Internet Co., Ltd, JasTel Network International). These sites include news outlets (nypost.com and dailymail.co.uk), wikileaks.org, the sites of circumvention tools (such as hotspotshield.com), and pornography. The fact that these sites were not blocked across all networks leads us to believe that Thai service providers may be filtering content based on broad government orders to block content that is deemed to violate lese majeste rules. On a positive note, WhatsApp, Facebook Messenger, and the Tor network appeared to be accessible across 16 networks throughout the testing period.
Given the limited transparency around information controls in Thailand and the potential implications they may have on human rights, we encourage ISPs to disclose their motivation and justification behind the blocking of sites and services. We also encourage public debate based on the findings of this study around the necessity and proportionality of information controls.
We thank the Open Technology Fund (OTF) and Access Now for funding this research. We also thank all the anonymous and brave volunteers in Thailand who have run and continue to run ooniprobe, thus making this research possible.
Note: This report was updated on 20th March 2017, following its publication.