[Special Feature: Social Media and Society] Takayuki Matsuo: Practice of Social Media Legal Affairs—Taking Account Suspension Responses under the Information Platform Act as an Example

The author has written works such as "Theory and Practice of Defamation on the Internet in Recent Precedents *1" and practices social media legal affairs, also known as SNS legal affairs. Social media legal affairs cover various topics; for example, SNS elections, deepfakes, and automated posting accounts called "bots" have recently become hot topics. However, this article focuses on account suspension *2.

In social media legal affairs, there are broadly three perspectives: the subject, the expresser, and the operator.

The subject is often called the "victim," referring to the person who claims to have suffered damage due to an expression on social media. While they are indeed often victims, legal claims may not always be established in the end; therefore, this article uses the neutral term "subject." Subjects seek post deletion, disclosure of the sender, apologies, and prevention of recurrence (including account suspension).

The expresser is often called the "user" or "sender," referring to the person who engages in activities such as posting or broadcasting on social media. From the subject's perspective, they may be called the "perpetrator." Expressers generally do not want their posts deleted by the operator (though they often delete posts themselves if they are indefensible). Furthermore, from a privacy standpoint, they do not want their anonymous accounts to be unmasked. Additionally, since they treat that social media as their "place of belonging," they strongly reject (unjust) account suspension, which means being deprived of that place.

The operator, sometimes called the social media operator or platform operator, is the entity that runs the social media in question. Operators receive requests for disclosure and deletion (account suspension) from subjects, while expressers demand that they do not disclose information, delete posts, or suspend accounts. They are truly in a position of being "caught in the middle."

As a lawyer, I have handled practice from all of these perspectives.

Here, if the service provided by the operator becomes a "lawless social media environment," it will have negative effects such as users leaving. Therefore, operators want to perform content moderation, such as deletion, for certain problematic posts. However, as mentioned, excessive deletion, disclosure, or account suspension leads to strong protests from expressers. Conversely, insufficient action leads to strong protests from subjects. In this situation, the former Provider Liability Limitation Act established requirements for deletion and disclosure by operators, ensuring that operators would not be held liable as long as they handled content moderation according to those requirements.

Taking deletion as an example, unless there were "reasonable grounds to believe that the operator could have known that the rights of others were being infringed by the distribution of information" (Article 3, Paragraph 1, Item 2 of the Act), the operator was not liable to the subject for distributing the post. Furthermore, even if they actually deleted it, they were not liable to the expresser if there were "reasonable grounds to believe that the rights of others were being unjustly infringed by the distribution of information" (Paragraph 2, Item 1 of the same Article) *3.

In short, a system already existed before the amendment to the Information Platform Act: if a subject claimed a post was illegal and requested deletion, and the operator had reasonable grounds to believe it was an infringing post, the operator could delete it without fear of being held liable by the expresser.

As stated, while I believe that a certain level of content moderation itself is necessary, the method used can significantly impact the rights of the expresser. Operators possess various means of content moderation, such as deleting individual posts, temporary account suspension, and (temporary) functional restrictions.

Among these many means, "(permanent) account suspension" has an extremely large negative impact on the expresser. In particular, terms of use often prohibit those who have received a permanent suspension from creating another account thereafter. In that sense, suspension is an extremely heavy penalty that prohibits future expression on that social media. For example, even if an expresser actually violates the terms of use and the terms allow for suspension, a minor violation should not immediately justify suspension. Such a penalty should be limited to cases where the measure is justified in comparison with other options—specifically, when milder measures cannot achieve the purpose of making the expresser comply with the terms or protecting the rights of the parties involved. Despite this, the issue of some operators choosing suspension relatively easily has been raised for a long time. For instance, in the 2024 Hyogo Prefecture gubernatorial election, the X account of Governor Saito's opponent was suspended multiple times during the election period. It has been pointed out that X's AI algorithm may have been "tricked" by organized reporting, leading to repeated suspensions *4.

A major problem is that some operators have terms of use or guidelines that do not clearly define the criteria for choosing the final resort of permanent suspension, which can be read as allowing arbitrary suspension at the operator's "discretion." In principle, suspension is a breach of contract by the operator regarding the social media service; for such a breach to be justified, strict requirements must be met. Even if defined in the terms of use, allowing such penalties to be carried out arbitrarily is unacceptable given the importance of social media services and the social responsibility of operators.

I have fought against operators on behalf of expressers for many years regarding inappropriate suspensions, including the unfair use of AI algorithms. For example, I represented a VTuber in a lawsuit against an international video streaming platform at the Tokyo District Court and won *5.

In such practical responses, I have achieved the lifting of many account suspensions through a two-step approach: first attempting to resolve the suspension through out-of-court communication, and if that is impossible, proceeding with legal action. That is, if an unfair suspension is simply an error by an AI algorithm, a notice from a lawyer acting as a representative may prompt a human staff member to review it, realize the error, and lift the suspension. If the suspension is not lifted after the notice, using legal procedures allows a Japanese lawyer to investigate which post was judged as a serious violation requiring suspension and why, which may lead to the suspension being lifted.

As mentioned above, I understand the necessity of content moderation itself and do not argue that "no content moderation should be done at all" *6. However, since content moderation methods range from high to low impact on the expresser, the choice should be appropriate to the degree of the expresser's actions.

I also understand the necessity of using AI for such content moderation. However, problems can arise, such as algorithms becoming unfair *7. Therefore, to prevent problematic processing by AI from being presented to the expresser as the company's official judgment, responses including a human-in-the-loop process, such as the so-called "Centaur model," should be considered *8.

The Information Platform Act expects problematic posts to be deleted quickly outside of court proceedings by mandating that certain large-scale operators establish systems for deletion and other actions.

Specifically, certain large-scale operators are designated and subject to special obligations based on three requirements: scale requirements such as "10 million or more monthly active users or 2 million or more monthly posts," the technical feasibility of measures to prevent the transmission of infringing information, and not being a specific telecommunications service with a low risk of rights infringement. Currently, designated entities include Google (YouTube), LINE Yahoo (Yahoo! Chiebukuro, LINE OpenChat, etc.), Meta (Facebook, Instagram), TikTok Pte. Ltd. (TikTok, TikTok Lite), X Corp. (X), Dwango (Niconico), CyberAgent (Abema Blog), Shonan Seibu Home (Bakusai), and Pinterest Europe Limited (Pinterest).

Under the Information Platform Act, these large-scale operators are primarily required to respond to rights infringements on their platforms as follows:

• Publication of methods for accepting deletion requests (Article 22): They must establish and publish online methods for accepting deletion requests from subjects. These methods must allow for requests in Japanese and must not impose an excessive burden on the subject.

• Conduct of infringement information investigations (Article 23): Upon receiving a deletion request, they must conduct the necessary investigation without delay.

• Appointment of specialists (Article 24): They must appoint specialists (such as lawyers) with sufficient knowledge and experience in handling rights infringements to conduct necessary investigations without delay upon receiving a deletion request.

• Notification of results to the subject (Article 25): They are obligated to notify the subject of whether deletion measures were taken and the reasons why, in principle within 7 days of the request. If no measures are taken, specific reasons must be provided to assist with any subsequent requests.

• Publication of deletion criteria (Article 26): They must publish their own deletion criteria and take deletion measures in accordance with them.

• Notification to the expresser (Article 27): If deletion measures are taken, they must notify the sender of that fact and the reasons without delay.

• Publication of implementation status (Article 28): Once a year, they are obligated to publish their operational status, including the status of deletion requests received, notifications to subjects and expressers, and their own evaluations of these situations.

In case of violation of these obligations, the Minister of Internal Affairs and Communications can issue recommendations or orders to correct the violation (Article 30). Failure to comply with an order without a justifiable reason results in penalties such as imprisonment or fines for the individual, and fines for the corporation (Articles 35 and 37).

Interestingly, the Information Platform Act includes provisions regarding account suspension.

Specifically, Article 2, Item 8 of the Act defines "measures to prevent the transmission of infringing information," which refers to what is commonly known as deletion *9.

In contrast, Item 9 of the same Article defines "measures to prevent the transmission of said information and to stop the provision of specific telecommunications services to the sender of said information" as "service suspension measures." These service suspension measures involve stopping the provision of the service itself—in other words, not providing the service at all—and thus include the concept of account suspension. It should be noted that this is not limited to "permanent" suspension; temporary suspension is also conceptually included in service suspension measures.

Furthermore, Article 26, Paragraph 2, Item 2 of the Act establishes a duty of effort for large-scale operators when setting criteria for deletion and other actions: "In cases where service suspension measures may be taken, the criteria for implementing service suspension measures shall be defined as specifically as possible." *10

This applies only to large-scale operators and is merely a duty of effort. Nevertheless, it has the potential to be a more effective response to the current problems mentioned above, such as unclear suspension criteria or terms of use and community guidelines that allow for discretionary suspension.

This can be seen as an indirect provision that account suspension, as a final resort, should be implemented only when it is truly unavoidable and based on criteria defined specifically in advance. I believe this is very suggestive for future account suspension practice.

As described, the Information Platform Act has introduced provisions pointing in a desirable direction regarding account suspension.

However, as mentioned in Section 3, defining specific criteria for account suspension remains a duty of effort. Furthermore, at least in terms of wording, it only states to "define as specifically as possible" and does not specify the concrete content that should be defined.

Regarding these challenges, for example, as part of the publication of implementation status mentioned in Section 2, Article 18, Paragraph 5, Item 8 of the Enforcement Regulations of the Information Platform Act requires the publication of the number of account suspensions categorized by reason and the circumstances leading to the measures. Additionally, Item 10 requires the number of suspensions using AI, Item 15 the number of appeals against suspensions (Item 16 for those using AI), and Item 17 the number of withdrawals of suspension measures (Item 18 for those using AI). I hope that the Information Platform Act and its practice will further improve based on the content of these disclosures.

I have briefly discussed social media practice from the perspective of account suspension response. While some individual details may vary, the intent of the Information Platform Act itself is very good. I strongly hope that future operations, especially regarding account suspension, will accumulate practices that fully reflect this intent.