Intermediary liability: Not Just Backward but Going Back[1] 

*This paper has been made part of Harvard University Berkman Center’s publication, “Governance of Online Intermediaries: Observations from a Series of National Case Studies.”


PARK Kyung-Sin (“K.S.”),

Professor, Korea University Law School;


Executive Director, PSPD Law Center

Twitter: @unbeatenpath


*This is the result of collaboration with Harvard University’s Berkman Center for Law and Society.




This paper provides an analysis of the Korean “Act Regarding Promotion of Use of Information Communication Networks and Protection of Information” that governs intermediary liability in Korea for defamatory or otherwise rights-infringing contents.  This study makes the case that the Act’s Article 44, 2, which should have created protection from liability like other intermediary liability regimes around the world, has become instead an affirmative way to impose intermediary liability.  The paper also gives an overview of relevant court cases, the latest (2009) of which, in author’s analysis, has had a “crushing” effect on protections from liability because it imposed a liability for contents that the intermediary was not aware of and was not give any notice of; and gives evidence for the argument that Article 44, 2 is unconstitutional because it imposes on intermediaries an de jure or de facto obligation to take down lawful contents.  Citing statistics on compliance with take-down requests by the three major intermediaries in the country, the author observes, as a result of such liability-imposing regime, that the sheer volume of censorship has become problematic, and that politicians use requests to take down legal content that is critical of their policy decisions- such as a posting criticizing Seoul City Mayor’s ban on assemblies in Seoul Square.

The Korean case provides a lesson for intermediary liability rules in general that what seems innocent is not really so.  To preserve the future of the Internet, an innocent sounding rule that an intermediary should be responsible for removing unlawful content should be checked for wisdom.  However, Korea shows a country where the special characteristics of the Internet are taken into account only in a direction suffocating the power of the Internet.


Table of Contents

1. Introduction

2. Landscape of Korean Intermediaries

3. Korea’s Intermediary Liability Regime  

a. Intermediary liability in general

b. Korean law: liability-exempting or liability-imposing

c. De Facto Interpretation: on-demand takedown obligation

d. Intermediary liability in court

4. Result: Private Censorship

5. People’s Response: Constitutional Challenges

6. Conclusion and (Impact Assessment)


  1. Introduction


This paper surveys South Korea’s landscape for intermediaries, analyzes the regulations thereon, and their impact on the society in general in response to the Network of Center’s Guiding Questions on the Online Intermediaries research project.  The author provides an overview of the country in section 2; discusses Korea’s intermediary liability regime in section 3; presents their impact on the industry and society in section 4; follows how law and industry have dynamically interacted in section 5; and finally concludes with suggestions for the future.


  1. Landscape of Korean intermediaries

 a. Market survey


As of 2013, Korea had a total population of about 48 million people (83% urban) with an Internet penetration rate of 84%, mobile penetration rate of 110%[1], mobile Internet penetration rate of 75%, and Facebook penetration of 27%[2].  See below for comparison to Japan, U.S., and the world average.



Korea Japan US World average
Population 48 million 127 million 312 million
Internet penetration 84% 79% 80% 52%
Mobile penetration 110% 109% 103% 93%
Internet mobile penetration 75% 48% 60% 21%
Facebook penetration 27% 17% 56%


Korea’s major intermediaries for each type are as follows:


  • Search engines: NAVER, the local portal, has maintained 73% market share. DAUM, the second largest local portal has about 21%, with GOOGLE covering the small remainder 3% (December 2012).[3]


  • Micro-blogging: TWITTER almost monopolizes the market, but if you include non-micro blogging, NAVER still covers 80% of domestic bloggers.[4]  In 2007, NAVER already topped user visits per month[5] and its dominance grew over time ever since.


  • Social Media: 31.3% of all people use SNS (increase by 7.8% in 2013, fast-growing), who use in order of the number of accounts, KAKAO STORY (SK Communications) 5.5%, etc. as of January 2014[8].  However, I personally think that the Kakao Story numbers are exaggerated by the users who were given the Story accounts by default due to their membership with Kakao Talk, the dominant private messaging services, not really a social “networking” service where past postings remain online for future “friends” to see.  Weighing the time spent using the services, I believe that FACEBOOK is by far the most widely used social networking service in Korea.  This is quite a change since 2010 when Cyworld accounted for 50% of social media users.[9]  Leaving out the messaging services, the rankings are as follows:


South Korea SNS 2014 : Own an Account (Monthly Active User)

Any SNS 84% (48%)

Facebook 75% (36%)

Twitter 56% (22%)

Google+ 38% (7%)

Me2Day 33% (7%)[10]


  • Private messaging: Kakao Talk almost monopolizes the market.[11]


  • User Created Content: YouTube has 75% but only in video contents.[12]


  • Platform: Google Play 75.2%, due to the dominance of Samsung (100% Android) in phone markets (Apple 17.9%, Blackberry and Windows each 4%)[13]


As part of the overall Internet economy, the mobile Internet is most often used for search (96.8%), and second most for SNS (50.4%), shopping (36.4%), banking (33.1%), etc.  Time-weighted, it is used most for chatting (81.2%), phone calls (visual incl.) (69.7%), texting (69.%), and searches (42.8%).[14]


As expected, the top uses of mobile Internet are not typical revenue-generators.  Below are the revenues of top 10 Internet companies in Korea:


Top 10 Internet companies (by revenue)

Naver (2.3 billion USD)

Nexon (1.6 billion USD)

NCSoft  (750 million USD)

NHN Entertainment (640 M)

eBay(640 M)

Daum (530 M)

Net Marle (497 M)

Neo Wiz (443 M)

Smilegate (360 M)

Wemade (227M)[15]


Notice, out of 10 companies, the majority are game companies.  Only NAVER and DAUM are portals.  FACEBOOK, TWITTER (SNS), KAKAO are not major revenue-generators.  Google Play revenues are not significant, either.


b. Social significance of different intermediaries

In non-economic terms, certain intermediaries are more relevant than others – e.g. in terms of market share, popularity, usage patterns, and their impact on society.  NAVER and DAUM curate and present other agencies news in their own pages, host original user-created discussion pages, and blogs (NAVER) and cafe pages (DAUM), which have become major platforms for political debates.   FACEBOOK has become the socializing platform of choice for both conservative and progressive circles.  TWITTER, which had become the main battleground for political discussions, became even more famous as it was later revealed that National Intelligence Services, the country’s intelligence agency, had conducted major public-opinion-manipulation campaigns using TWITTER before and during the Presidential election period in 2012.[16]


In late 2014, the Korean intermediary, KAKAO TALK, the dominant messenger service provider, became the center of public attention when the Prosecutors’ Office announced a new campaign to track down and indict the postings “causing division in national unity and skepticism of the government” for criminal defamation, and in doing so, mentioned KAKAO TALK as a possible target for such search and seizure.  This shocked the entire nation, 90% of who use KAKAO TALK because it has been a private messenger service connecting only those who know each other, and many ‘migrated’ to a foreign service TELEGRAM, whose server is located overseas, apparently safe from Korean authorities’ search and seizure.[17]

c. State paternalism

Indeed, one significant factor affecting online intermediaries is state paternalism, which pervades the country’s industrial institutions and practices.  For instance, all Internet companies with capital larger than about USD 100K are required to register and are given a “value-added telecommunication business” number, which can be taken away if they do not operate in compliance with the government’s laws and regulations or their operation “significantly hurts consumers’ interests”.[18]  This environment creates a cloud under which the domestic companies feel the pressure to comply with even extra-legal guidance of the government.  For instance, as you will read below, the “temporary take-down” regulation can be read to be only optional but effectively works as if it is mandatory, as do several other “optional” regulations such as Korea Communication Standards Commission’s “correction requests (to take down contents)”[19] and warrantless subscriber data requests, the compliance rates of which were respectively near 100% until a huge judgment came down on the latter in October 2012 in a consumer lawsuit filed by PSPD Law Center.[20]

d. Foreign companies

The regulations, hard and soft, apply equally to FACEBOOK, TWITTER, GOOGLE, and MICROSOFT, which all have local offices but whose servers are located overseas, exempting the owners from local income tax liabilities.  The extraterritoriality of the servers also has provided a rationalization for the fact that the government has not applied various intermediary regulations to these overseas providers, creating what domestic competitors decry as “reverse-discrimination”.[21]  The most infamous domestic-only regulation was mandatory identity verification rule, which was snubbed only by overseas providers before it was struck down in 2012 in a constitutional challenge filed by PSPD Law Center.[22]


3. Korea’s Intermediary Liability Regime

a. Intermediary liability in general

What defines the Internet?  The defining feature of the Internet is its nature as an extremely distributed communication platform, so distributed that it allows almost all individuals to participate in mass-scale communication.  All individuals are allowed to post individual views and opinions without anyone’s approval, and all individuals are allowed to view and download all other individuals’ postings.

How some people react to questionable material found online shows how they have not accustomed themselves to this freedom of the Internet.  They think that Internet companies should be responsible for contents on their services.  Yes, illegal activities such as defamation and copyright infringement that abuse the power of the Internet should be combated.  However, unless we want to paralyze the freedom of unapproved uploading and viewing and therefore the power of the Internet, an intermediary that cannot possibly know who posts what contents, should not be held responsible for defamation or copyright infringements committed via some contents on its services.  If intermediaries are held liable for these unknown contents, the intermediaries will have to protect themselves by constantly monitoring what gets posted on their services.  If that happens, one can say that, when a posting remains online, it remains online at the pleasure and tacit approval of the intermediary that saw the posting and did not block it.  The power of the Internet — the freedom to post and download unapproved — will be dead.

For the same reason, no country imposes – for instance – content liability on broadband providers.[23]  No common carrier will be in business if it is held liable for all the criminal conspiracies and deals taking place over their networks.  Now, the same reasoning should be extended to the providers of web applications that greatly facilitate the exchange of ideas and contents, i.e. “portals” and “search engines”.  The only difference with the common carriers will be that the Internet companies carry the unlawful contents on their servers while the telecoms serve the contents en route.  While some will surely abuse free space created by these intermediaries, holding the intermediaries liable merely for creating this space would be too threatening to the future of the Internet.  Along this line of thought, on non-copyright-related contents, the U.S. even went a little further by claiming that no “interactive computer service” shall be considered a speaker or a publisher of those contents.[24]

However, as to other areas, many believe that there must be a limit on the exemption that the intermediary enjoys:  the intermediary should not be immunized for the infringing content that it is aware of or is given notice of and yet refuses to remove.  Yet this idea of a limited liability regime is not satisfactory because the intermediaries always face a stronger incentive to take down individual contents than an incentive to keep them on.  The reason is that, firstly, they are massive content processors whose interest in individual contents are minuscule, and secondly, tort liability regimes around the world are usually such that the legal exposure for keeping a posting on (a malfeasance) is always greater than the legal exposure for not keeping it (a nonfeasance).

Therefrom, many countries decided to set up “safe-harbor” regimes where the intermediaries will be exempt from liability if they choose to follow certain clearly defined procedures aimed at abating unlawful content.  The most widely popular such regime is the notice-and-takedown regime,[25] whereby an intermediary is given an exemption from liability as long as it removes content of which it is given notice of the infringing nature by a rights holder.  Importantly, the notice-and-takedown safe harbor is not applicable to the contents that the intermediaries have actual knowledge of their illegality even before and without notice provided by a rightsholder or any other person.

b. Korean law: liability-exemption or liability-imposition?


In Korea, the idea that the intermediaries must be given exemption from liability in the way of safe harbor on the Internet appears to have been misinterpreted: what we have is not an intermediary liability exemption regime but intermediary liability imposition regime.  The relevant provisions are as follows:


The Act Regarding Promotion of Use of Information Communication Networks and Protection of Information, Article 44-2 (Request to Delete Information) reads:

Paragraph 1.  Anyone whose rights have been violated through invasion of privacy, defamation, etc., by information offered for disclosure to the general public through an information communication network may request the information communication service provider handling that information to delete the information or publish rebuttal thereto by certifying the fact of the violations.

Paragraph 2. The information communication service provider, upon receiving the request set forth in Section 1 shall immediately delete or temporarily blind, or take other necessary measures on, the information and immediately inform the author of the information and the applicant for deleting that information.  The service provider shall inform the users of the fact of having taken the necessary measures by posting on the related bulletin board.


Paragraph 4. In spite of the request set forth in Section 1, if the service provider finds it difficult to decide whether the rights have been violated or anticipates a dispute among the interested parties, the service provider may take a measure temporarily blocking access to the information (“temporary measure”, hereinafter), which may last up to 30 days


Paragraph 6. The service provider may reduce or exempt the damage liability by taking necessary actions set forth in Paragraph 2.

As is immediately apparent, the provision is structured not with such phrases as “the service provider shall not be liable when it removes . . .” but starts out with a phrase “the service provider shall remove …”.  Paragraph 6, referring to the “exemption from or reduction of liability in event of compliance with the aforesaid duties,” makes a feeble attempt to turn the provisions into an exemption provision like the notice-and-takedown of the Digital Millennium Copyright Act.  However, exemption here is not mandatory, but dependent on the courts deciding on the intermediary liability literally because the liability “may be reduced or exempted” not “shall be exempted”.  In fact, none of the service providers interpret Article 44-2 as an exemption provision that they are allowed to deviate from on the simple penalty of foregoing a safe-harbor.  All of them interpret it as an obligatory provision that they must comply with.

Indeed, historically, the predecessors of Article 44-2 (Article 44 Paragraphs 1 and 2 of the Network Act enacted 2001.7.16, Law No. 6360)[26] simply required the service provider to take down content upon the request of a party injured by that content and did not provide any exemption.  Article 44 began as a simple idea that the service provider shall be responsible at least for the content that is infringing if someone had complained about that content previously.  Then, many service providers complained that they were not capable of determining whether certain content is infringing or not.  In response, the law was amended in 2007 (Enacted 2007.7.27 Law No. 8289) into Article 44-2 to create a “temporary (blind) measure” for “border-line” content, which the service provider can now resort to in fulfilling their responsibility under the previous law.[27]  Together with that amendment, the noncommittal reference to possible “reduction or exemption” found its way into the law.  The central idea that continued to remain was that the intermediary must remove an infringing content upon demand.

The general idea of holding the intermediaries liable for identified infringing content seems innocuous but the Korean case shows compellingly why we should abandon it, as we shall see below.

c. On-Demand Takedown Obligations


As I will explain below, Article 44-2 Paragraphs 1, 2 and 4 of the Act Regarding Promotion of Use of Information Communication Network and Protection of Information (“Network Act”) is that they require the service provider to take at least a “temporary measure” on all contents upon which the takedown request has been made regardless of the legality of the content.

The first possible interpretation is that the statute sets up such on-demand takedown obligation explicitly.  Although it speaks of an obligation to remove only when someone “whose rights have been violated” makes such request, it is impossible to know ex ante whether rights-infringement has taken place.  So the only feasible interpretation is that such obligation arises whenever someone who thinks and proposes that his rights have been violated.  Going further on this line of interpretation, this obligation can be instead filled by “temporary measure”, too, but that is the minimum:  the intermediary should take some abatement action even against.  Now, the statute thus interpreted will be against all known constitutions and international human rights treaties which allow freedom of speech to be violated only to relieve some other rights or values.

Another more generous interpretation is possible:  As you can immediately see from Paragraph 1 and 2, if someone complains of their infringed rights, the provider must take down the content if it is infringing.  Now, there will be no problem if the takedown obligation applies only to those contents that actually injure others.  Indeed, Paragraph 1 limits its application only to “anyone whose rights have been violated.”

However, even if that is the case, the service providers will have a strong incentive for removing the content regardless because otherwise the provider must risk being found wrong by courts and therefore liable as a contributor to the dissemination of the infringing content.  Usually, the service providers retain editorial control over the content through their Terms of Service so that they will not be liable vis-à-vis the authors of the contents for removing the content.  Article 44-3 of the Network Act even codifies this rule.[28]  On balance, the service provider always has stronger incentives to take down than to keep it up.

Now, Paragraph 4 states, in paraphrasing, that “the service provider may take a temporary measure (instead of permanent removal), if it is difficult to know whether the contents are infringing or when a dispute is expected between the parties”.  This should mean that, even if the contents are later found to be infringing, the service provider will not be held liable for showing the content if it has taken a temporary measure.  It seems to soften the de facto censorship effects of Paragraphs 1 & 2 by providing a less drastic alternative to a permanent removal, but it does exactly the opposite:  What is diabolical is that the permissive “may” in Paragraph 4 will encourage the intermediaries further into removing perfectly lawful contents.  This further aggravates the imbalance of incentives in favor of restricting the content rather than keeping it up.

Make no mistake about it: under this second interpretation, the failure to take abatement action will result in liability only if the content is later found to be actually infringing.  However, the intermediaries not knowing for sure which ones are infringing will have strong incentives to take down even lawful content instead of risking being found liable later.  Maybe a better expression of the dilemma will be that the providers will be “chilled” into doing so, not because the concept “rights-infringing” is vague all the time but vague ex ante.  On top of that, Paragraph 4 provides yet another incentive in favor of removing the content by providing exemption from any liability for doing so.

Initially, the service providers were expected to gravitate away from permanent removals, for which there is no ex ante exemption, toward temporary measures, for which there is ex ante exemption.  This prediction turned out to be true.  NAVER, the number one content host, has often responded to all takedown requests with only temporary measures; DAUM, the number two content host, eventually caught up in 2010.

In sum, contrary to the spirit of intermediary liability regimes around the world aimed at shielding the creators of online space from liability for what goes on in that space, Korean law goes backward: it ends up imposing the de facto obligations on the intermediaries to censor lawful material, the obligations that did not exist when Article 44 or 44-2 were not legislated.  The next section examines how courts have dealt with intermediary liability before or without the current Article 44/44-2 regime.

d. Intermediary Liability in Court[29]


The Korean Supreme Court has ruled three times significantly on intermediary liability.   In 2001, the Court held an electronic bulletin board provider liable for refusing, even upon demands both by the injury claimant and a government censorship body, to take down for a period of for 5-6 months postings deprecating a pop singer’s fan.  The Court ruled that the intermediary had “a duty to take adequate measures when it knew or had reason to know of a defamatory posting.”[30]  Nothing unusual.

In 2003, when the Court was asked to find an intermediary liable for postings defaming a local politician, the Court took that as an opportunity to further limit when such duty to take adequate measures arises. [31] The Court held that, an intermediary, even if it knew or had reason to know of the defamatory material for 52 days, should not be held responsible unless a comprehensive analysis of the following factors point to such responsibility:  “the posting’s purpose, content, duration and method, the damages it has caused, the relationship between the speaker and the injury-claimant, the claimant’s attitude including whether rebuttal or takedown was requested, the size and nature of the site posted, the degree of for-profit nature of the site, when the operator knew or could have known the posting’s content, and the technological and pecuniary difficulty in taking down, etc.”[32]  Having said so, the Supreme Court reversed the lower court that imposed the liability for pre-takedown exposure.  The Supreme Court’s rather terse ruling sounds very generous, refusing to impose liability even upon knowledge of some indiscretion, especially when it was before the exemption provision was added to Article 44-2.  However, the ruling stands on the narrow fact that the intermediary here did comply immediately with the takedown request.  Some said it made sense to require knowledge of the illegal character of the content[33].

Then in 2009, a crushing judgment[34] came out where the Korean Supreme Court issued a decision holding web portal sites Naver, Daum, SK Communications, and Yahoo Korea liable for the defamation of the plaintiff when user postings on those sites accused him of deserting a girlfriend upon the second pregnancy after he had he talked her into aborting the first, after which the girlfriend then committed suicide.  The court upheld judgments of 10 million won, 7 million won, 8 million won, and 5 million won, respectively, against these services.”

Specifically, the court held that (in paraphrase):


Barring special circumstances, the intermediary shall be liable for illegal contents to the same extent as a news agency and therefore shall be liable when (1) the illegality of the content is clear;  (2) the provider was aware of the content; and (3) it is technically and financially possible to control the contents.  On top of the duty to take down such contents immediately, the intermediary has a duty to block similar postings later on.  The Court will find the provider’s requisite awareness under (2) above:

a) when the victim has requested specifically and individually for the takedown of the content;

b) when, even without such request, the provider was concretely aware of how and why the content was posted OR

c) when, even without request, it was apparently clear that the provider could have been aware of that content.

The end result is that the intermediary will be absolutely liable for a posting later found to be “clearly” defamatory if “it was apparently clear that the provider could have been aware of that content” even if the victim did not notify the intermediary of the existence of the content.

This sets up probably one of the most strict intermediary liability regime because it imposes liability for “unknown but could-have-known” contents.  Anupam Chander plainly describes this ruling as stating that a web service “must delete slanderous posts or block searches of offending posts, even if not requested to do so by the victim.”[35]  It is true that the intermediary may be held liable for the content that looks clearly illegal ex ante but even when the intermediary did not know?

True, DMCA notice-and-takedown immunity[36] does not apply to those contents that OSP had “actual knowledge” of the infringing nature or “awareness of facts or circumstances from which infringing activity is apparent.”  However, DMCA is a safe harbor provision.  It merely says that the safe harbor will not apply in case of “actual knowledge” or “awareness”.  It does not say that the OSP will be held liable in case of such knowledge or awareness.

Furthermore, there is a world of difference between possible awareness — meant by the phrase “could have been aware” in the Korean ruling — on one hand and “actual knowledge” or “awareness” on the other.  The intermediar n order to make sure that there are no unknown clearly defamatory postings that it “could have been aware” that they did not remove.  This sets up a general monitoring obligation that kills the power of the Internet.  Indeed, the Court does state that “[if the three conditions are met], the intermediary has a duty to take down such contents immediately AND block similar postings later.”

What is more, this was not even a case interpreting the Article 44/44-2 regime because the cases here concerned the intermediary’s role before or when the victim did not make a takedown request.  The Court was already ready to impose a publisher-like liability on the intermediary and a monitoring obligation.



  1. Result: Private Censorship


In summary, Article 44-2 states that all contents should be taken down upon demand even if lawful.  The Supreme Court decisions state that all unlawful contents should be taken down even if unknown to the intermediaries.  Article 44-2 and together encourage private censorship by the intermediaries.  On top of the censorship system triggered by private notices, Korean law provides for Korean Communication Standards Commission which issues “correction requests” to all intermediaries, including telecoms, to take down or block domestically the content the Commission found to be illegal.  What is significant for now is that these injunctive functions, together with monetary damages, anticipated by the above-described liability regime, will provide stronger incentives to the intermediaries to take a heavy-handed approach toward censorship.[37]  We will now look at some numbers and cases for illustration.

We will not look at copyright-related takedown notices, which may make up more than 90% of takedown requests in other countries, because the Korean Copyright Act sets up a different liability scheme for copyright-related takedown requests.  The Network Act’s liability scheme affects only takedown requests related to defamation, privacy, interference with business, etc.  Although the Network Act’s liability scheme on its face covers copyright as well, the Copyright Act’s scheme takes precedence in copyright issues in accordance with the principle of generalia specialibus non derogant.  Although there will be issues with copyright-related on-demand takedowns, the Copyright Act’s liability scheme has been quite similar to the American DMCA and now more so under the KORUS FTA-triggered amendment that closed the final loophole by making the liability exemption mandatory.

There is nothing similar to the Transparency Reports of U.S. OSP’s that is published by Korean intermediaries.  There are only statistics occasionally obtained through private sources along with legislators who exercise their clout through the agencies that can in turn make various disclosure demands on the intermediaries licensed or registered with them.  MP Choi Moon-soon obtained the relevant data from the top three top content host intermediaries though Korea Communications Commission and made the following disclosure in November 2010[38]:


<<Non-copyright-related Takedowns Pursuant to Article 44-2>>

OperatorsYears NAVER DAUM NATE Total
2008 31,953 27,454 691 60,098
2009 37,342 57,712 1,449 96,503
2010 up to September(estimated year-end figures) 27,914(37,125) 45,798(60,911)


956(1274) 74,668(99,310)


After learning that the number of takedowns executed by the top two content hosts exceeds that of other hosts greatly, MP Nam Kyung-pil obtained similar data on the two content hosts in October 2012[39] shown below.


<<Non-copyright-related Takedowns Pursuant to Article 44-2>>

OperatorsYears NAVER DAUM
2008 70,401 21,546
2009 83,548 50,860
2010 85,573 58,168
2011 123,079 86,431
2012 until July 104,578 40,538


Although the differences in the two tables need some explanations[40], we can make the following points of fact, uncontested:

  1. The number of URL takedowns privately requested under Article 44-2 of the Network Act for non-copyright purposes has increased over time.
  2. The annual number of URLs taken down by NAVER hover above 100,000 and that for DAUM is about 50-70% of NAVERs.


How serious is this?  There is nothing here that we can compare to the situation in the U.S. because Section 230 of CDA insulates the intermediaries from liability for defamation and other non-copyright related laws.  However, we can compare these Korean numbers to government-originating takedowns in other countries.  Google received only about 4,000 takedown requests in 2012 from the whole world, only about half of which Google complied with.[41] So, 100,000 in Korea vs. 2,000 the whole world vis-à-vis Google! As another example, the Korean government’s censorship body, the Korean Communication Standards Commission, issued 54,385 takedown requests to various intermediaries in 2011, out of which only 668 were related to defamation and other rights infringement.[42]  Although the number of URLs is usually greater than the number of requests — for each request may cover more than one URL — the rights-infringement category of KCSC activities usually covers less than 10 URLs.  It means that private censorship through Article 44-2 is underlying more than 10 times the number of rights-infringement takedowns executed by the Korean government.

It is not just the volume of censorship that is problematic.  Politicians and government officials often make the takedown requests on postings critical of their policy decisions that are clearly lawful as illustrated below:

  • A posting[43] critical of a Seoul City mayor’s ban on assemblies in the Seoul Square
  • A posting[44] critical of a legislator’s drinking habits and introducing his social media account;
  • Clips of a television news report on Seoul Police Chief’s brother who allegedly runs an illegal brothel-hotel;[45]
  • A posting critical of politicians’ pejorative remarks on the recent deaths of squatters and police officers in a redevelopment dispute[46]
  • A posting calling for immunity from criminal prosecutions and civil damage suits on labor strikes.[47]
  • A posting by an opposition party legislator questioning a conservative media executive’s involvement in a sex exploitation scandal related to an actress and her suicide.[48]


  1. People’s Response: Constitutional Challenge


It is okay not to institute intermediary immunity regimes such as the United States’ CDA Section 230 or DMCA Section 512 that shields intermediaries from liability for even unlawful contents.  However, Korea does much worse:  It chills the intermediaries into taking down even lawful content as evidenced by the examples above. The PSPD Public Interest Law Center and others filed a constitutional challenge against Article 44-2 of the Network Act on the theory that the total result of the aforesaid provisions is that “Thou Shall Delay Saying What Others Dislike, As Long As 30 days.”[49]  For the Constitution does not authorize abating a speech not violating others’ rights, the aforesaid provisions de facto requiring even lawful contents to be abated for up to 30 days therefore are unconstitutional.

Under the current statutory scheme, the temporary removal can be up to 30 days. DAUM set it at the maximum of 30 days, while NAVER set it at a period lasting until the publisher requests reposting. NAVER’s system looks a lot like a notice-and-takedown without mandatory exemption.  However, the statute is requiring even NAVER to take down what is clearly lawful at least once. The rule “Thou Shall Not Say What Others Dislike Unless Thou Have Courage to Say Twice” is equally unconstitutional.

In 2012, the Constitutional Court rejected the challenges as follows:[50]


The instant provisions are purported to prevent indiscriminate circulation of the information defaming or infringing privacy and other rights of another, and therefore have a legitimate purpose. . . Temporary blocking of the circulation or diffusion of the information that has the possibility of such infringement is an appropriate means to accomplish the purpose.


Freedom of speech requires absence of restriction in form, method, and timing of speech.  Especially, in relation to publishing one’s opinions on a certain issue or event, the ‘temporal pertinence ‘ i.e. making a remark appropriate to the event in a time proximately related to the subject of that opinion is an important component of free speech and should be maximally guaranteed.  This is an important function of freedom of speech that calls for self-correction through rebuttal and discussions about that speech, conducted at ‘the marketplace of ideas.’  Therefore, the instant provisions’ ‘temporary measure’ depriving the speech of the temporal pertinence by blocking access through information communication network presents a grave restriction on free speech.


However, . . .when another’s personal rights such as privacy or reputation are infringed or are anticipated to be infringed, a need to temporarily block the infringing information is greater than the need to guarantee the temporal pertinence of the information.  The fact that the content was disclosed may be further propagated through other means, and may cause privacy-infringement and defamation to an equal extent.  In such situation, publishing a rebuttal by the infringement complainant, blocking of the links, search restrictions, expeditious dispute resolution, etc., cannot be effective alternatives to accomplish the legislative purpose.


When a temporary measure is taken for the reason that “it is difficult to judge whether the rights have been infringed or when a dispute between the interested parties is anticipated”, the degree of restriction on the poster’s freedom of speech becomes greater. . . .However, in this situation, such measure has the effect of preventing frivolous improvised attacks or the spreading of information that as a result infringe on another’s rights in anonymous cyberspace.  . .

What was encouraging was that the Constitutional Court saw through to the practical effects of the provisions and recognized that the provisions are in fact tantamount to requiring the takedown of contents that are not illegal.  The Court itself states: “if the prerequisites are met, the service provider must without hesitation take the temporary measure.”

However, the opinion takes a curious turn and rationalizes the blocking of content on the basis of a mere “anticipation” of infringement.  That speech can be banned on the basis of a possible illegality is a far departure from the established rules of free speech such as a clear and present doctrine, void-for-vagueness, prior restraint ban, etc.  The reason for such leniency is found in the earlier portions of the decision emphasizing how fast, far, and wide defamatory information travels through the Internet. However, the decision does not mention how fast, far and wide corrective information can travel. Sure, the Internet’s self-corrective nature cannot be the basis for exempting all unlawful activities on the Internet.  However, communicative efficiency of a medium cannot be a justification for taking down contents that are lawful on that medium.

In all other media, only proven illegality can form the basis of liability, intermediary or primary.  The Korean intermediary liability regime will impose liability for only a provisional illegality if it takes place on the Internet.  This constitutes discrimination against the Internet as a medium.  Now, it is not a frivolous question how humanity should deal with the special characteristics of the Internet, which calls for more research.


  1. Conclusion and Impact Assessment


The Korean liability regime starts out with an innocent-sounding rule that an intermediary shall remove any user-created content infringing on the rights of another.  The regime adds yet another innocent-sounding rule that an intermediary is free to remove a UCC temporarily as long as the intermediary anticipates a dispute or faces difficulty in deciding on the lawfulness.  Such regime, exempting not posting but only removing of a post, has caused in Korea the natural result of rampant private censorship in the country, taking down much content duly informative to the public on civic affairs.  The courts have not behaved better, imposing liabilities on the intermediaries for not taking down unknown content for which there was not even a takedown request.  Civil society has responded with a constitutional challenge, which ended with a surprising decision by the Constitutional Court that the Internet, due to its hyper-efficient mediating power, must be discriminated against so that even lawful content must be subject to temporary removal if there are people who alleged an injury.

[1] “Unconstitutionality of Korea`s Temporary Blinds on Internet -“Thou Shall Not Speak for 30 days What Others Do Not Like”, Joongang Law Review, Vol.11 No.3 Pages 7-51 [2009]

[2] We Are Social Singapore, “Global Digital Statistics 2014”, January 2014, page 146-146 (cited sources: ITU, Facebook, U.S. Census Bureau, Global Webindex)

[3] Ministry of Science, ICT and Future Planning and Korea Internet & Security Agency, 2013 Korea Internet White Paper,, p. 180

[4] There does not seem to be statistics tracking  blogging or micro-blogging separately. “80%” is usually tossed around by Internet pundits who seem to derive that number from the search engine market share for the reason that bloggers are likely to expect search engines to promote the blogs on their own services and therefore likely to use the blog platform affiliated with the most popular search engine NAVER.

[5] Nielson Korean Click Co., Ltd., “Domestic Blogging Services: Growth and Change”, November 14, 2007

[6] An Instagram-like SNS launched by Kakao Talk, the dominant private messaging service, opened

[7] A My Space-like service launched by the SK conglomerate.  This remains the only non-telco intermediary founded by Korean chaebols.

[8] Korea Information Society Development Institute, KISDI Stat Report “SNS Usage Analysis(SNS 이용추이 분석)” (2013.12.26)

[9] Ministry of Culture, Sports and Tourism,

[10] We Are Social Singapore, infra., p. 148

[11] Newsis, “Kakao Talk’s Market Share at 92%. . Bandwagon Effects in Mobile Messenger Service, Twice That of Mobile Telecom”, September 23, 2014,

[12]  Newsis, “YouTube, Clearing the Video Market Thanks to Mandatory Identification Rule”, October 9, 2013,

[13] 2013 Korea Internet White Paper, infra, p. 29,

[14] Korea Internet and Security Agency, “Year 2013 Mobile Internet Usage Survey (모바일인터넷이용실태조사)”, January 15, 2014.

[15] Blog ‘Under the Radar’, “2013 Internet Industry, Top 10 Revenue Generators”, March 7, 2014,

[16] New York Times, “Prosecutors Detail Attempt to Sway South Koran Election”, November 21, 2013.

[17] BBC “Why South Koreans are Fleeing the Country’s Biggest Social Network”, October 10, 2014.

[18] Article 27 Paragraph 2 of the Telecommunications Business Act

[19] See K.S. Park, “Administrative Censorship on Internet in Korea”,

[20] See K.S. Park, “Internet Surveillance in Korea 2014”,  I myself had the fortune of initiating and directing the legal campaign for the lawsuit, which is now pending in the Supreme Court.

[21] Business Korea, “Korean ICT Companies Suffering from Reverse Discrimination due to Governmental Regulations”,

[22] Constitutional Court’s Decision 2010 Hunma 47, 252 (consolidated) announced August 28, 2012.  K.S. Park, “Korean Internet Identity Verification Rule Struck Down”

[23] Section 512 (a) of the Digital Millennium Copyright Act

[24] Communications Decency Act of 1996: 47 USC 230 “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”

[25] DMCA section 512 (c) and (g)



[28]  Article 44-3  The service provider may take a temporary measure voluntarily if it is recognized that the information circulated through the network operated and managed by the provider is infringing another’s rights.

[29] Woo Ji-Suk, “A Critical Analysis of the Practical Applicability and Implication of the

Korean Supreme Court Case on the ISP Liability of Defamation” LAW & TECHNOLOGY, Vol.5, No.4: pp78-98. July 2009 <>

[30] Supreme Court, 2001.9.7 Judgment, 2001Da36801

[31] Supreme Court 2003.6.27 Judgment, 2002Da72194

[32] Supreme Court, 2003.6.27 Judgment, 2002Da72194

[33] Hwang Sung-Gi,

[34] Please review a foreign scholar’s response to this ruling.  Anupam Chander, “How Law Made Silicon Valley” EMORY LAW JOURNAL [Vol. 63:639 (2014),

[35] Supreme Court, 2008Da53812, Apr. 16, 2009 (S. Kor.).

[36] Section 512(c)(1)(A)(ii), 512(d)(1)(A)

[37] Park, Ahran. “Internet Service Provider’s Liability for Defamation: South Korea’s Balancing of Free Speech with Reputation” Paper presented at the annual meeting of the Association for Education in Journalism and Mass Communication, The Denver Sheraton, Denver, CO, Aug 04, 2010 <Not Available>. 2014-01-06 <>




[40] NAVER’s numbers in the first table represent the number of requests, which can cover more than one URL, while the NAVER numbers in the second table represent the number of URLs taken down.  DAUM’s numbers in the first table include both permanent removals and temporary measures, i.e., blinds while DAUM’s numbers in the second include only temporary measures.  DAUM’s numbers in the second table more and more came to represent the total number of takedowns as DAUM cancelled its policy of undoing the blinds after 30 days, i.e., all temporary measures became permanent.




[44] The original posting now taken down is shown here.






[49] Park Kyung-sin, “Unconstitutionality of Korea’s Temporary Blinds on Internet – “Thou Shall Not Speak for 30 days What Others Do Not Like”, Chung-Ang-Bub-Hak (Korean) <>


[50] Constitutional Court 2012.5.31 Decision 2010 Hun-ma 88