Online Reputation Governance Monthly Report (August 2026)

Three Key Takeaways This Month

1. Regulatory Signals This Month: On August 4, the Cyberspace Administration of China (CAC) publicly exposed a batch of typical cases from the “Qinglang · Optimising the Online Business Environment — Rectifying Malicious Hyping of Enterprise-related Information” special campaign; four categories of accounts — those maliciously smearing and disparaging enterprises, hyping negative enterprise-related information, demanding money in the name of deleting posts, and conducting false reviews — were closed or otherwise dealt with in accordance with laws and platform agreements. On August 6, the Cybersecurity Review Office announced the launch of a cybersecurity review of products sold in China by Palo Alto Networks (派拓公司).
2. Case Insight This Month: An enterprise facing malicious smears is not confined to passive endurance — this month, the same car-review account received conclusions along two lines at once, judicial and administrative: the court held that its conduct constituted commercial disparagement, and the CAC’s notice reported that its account had been closed in accordance with laws and platform agreements. Yet in the same month, the “Zhu Zhiliao” (竹知了, bamboo cicada) incident also showed that once an enforcement action launched by an enterprise itself lacks explainable messaging, it can be rewritten into a different story in the court of public opinion.
3. Counsel’s Recommendation This Month: The legal legitimacy of rights enforcement and its online reputation safety are two independent lines of judgment — the former depends on evidence and statutory provisions, the latter on the precision with which complaint targets are screened and on the explainability of the action. The legal risk of batch complaints lies not in “complaining” as such, but in the granularity of the evidence item by item: under Article 1195, Paragraph 3 of the Civil Code of the People’s Republic of China, a rights holder who causes damage to a network user or network service provider through an erroneous notice shall bear tort liability.

Governance Signals This Month

I. The CAC Releases Typical Cases from the “Qinglang · Optimising the Online Business Environment — Rectifying Malicious Hyping of Enterprise-related Information” Special Campaign

On May 26, 2026, the Secretariat of the Office of the Central Cyberspace Affairs Commission issued the Notice on Launching the “Qinglang · Optimising the Online Business Environment — Rectifying Malicious Hyping of Enterprise-related Information” Special Campaign, deciding to launch, with immediate effect, a two-month special campaign nationwide. Under the work objectives set by the Notice, the campaign is to concentrate on clearing enterprise-infringing information and related accounts and to urge websites and platforms to build content management mechanisms for enterprise-related information, with the online business environment as its end point. On August 4, 2026, the CAC publicly exposed a batch of typical cases investigated and dealt with under the campaign. According to the notice, the leads for dealing with these accounts came from letters sent by relevant departments and local cyberspace administrations, and the targets were illegal and non-compliant accounts infringing the lawful online rights and interests of enterprises and entrepreneurs.

The four cases made public correspond one-to-one to the four categories of prominent problems listed in the campaign Notice. The first involves three car-review accounts (including the Weibo and Douyin account “Dianche Xiaofei” (电车小飞)), which the notice found to have published misleading information, disparaged the product reputation and business reputation of an automobile enterprise, and profited from the resulting marketing and traffic diversion; the accounts concerned have been closed in accordance with laws and platform agreements, corresponding to the second category of problems in the Notice. The second involves a workplace-themed public account and a finance-themed Weibo account, which the notice found to have distorted the interpretation of business strategies, exaggerated revenue conditions, and talked down development prospects at moments such as enterprises’ listing and financing or the publication of financial statements; these accounts were likewise closed, corresponding to the scenario of hyping around important time points within the first category of problems. The third involves a cross-platform matrix account, which the notice found to have repeatedly rehashed old news about enterprises and attacked corporate executives, while its operator demanded money from enterprises and sought business cooperation in the name of deleting posts and eliminating impact; the account has been closed, corresponding to the third category of problems, the pursuit of illegal gains. The fourth involves five review accounts, which the notice found to have published false and untrue online review information, disparaged enterprises’ product quality and product reputation, and misled consumer perception; they have been dealt with in accordance with laws and platform agreements, corresponding to the scenario of false reviews within the first category of problems.

For corporate reputation governance, the value of this batch of notices lies in mapping out the reach of administrative action and marking the interface with civil enforcement. The four categories of conduct named mostly fall at the same time, on the civil side, under Article 12 of the Anti-Unfair Competition Law of the People’s Republic of China (2025 Revision) — a business operator shall not fabricate or disseminate, or instruct others to fabricate or disseminate, false or misleading information that damages the business reputation or product reputation of other business operators; under Article 28 of the same law, a violation of that article shall be ordered by the supervision and inspection authority to cease the illegal conduct and eliminate its effects, and shall be subject to a fine of RMB 100,000 to 1,000,000, or RMB 1 million to 5 million where the circumstances are serious. On the litigation path, Article 8 of the Provisions of the Supreme People’s Court on Several Issues Concerning the Application of Law in the Trial of Civil Dispute Cases Involving Infringement upon Personal Rights and Interests through Information Networks (the SPC Online Infringement Provisions) makes clear that where a network user or network service provider uses defamation, disparagement, or similar means to damage public trust in a business entity or lower the social evaluation of its products or services, and the business entity requests that it bear tort liability, the people’s court shall support the request in accordance with law. The paid post deletion involved in the third case is subject to a further, independent rule: Article 10 of the same judicial interpretation provides that where the infringed party reaches an agreement with an infringing network user or network service provider under which one party pays remuneration and the other provides services such as deletion, blocking, or link disconnection, the people’s court shall find the agreement invalid — an enterprise that, when faced with demands for “post deletion fees” or “article withdrawal fees,” pays in exchange for deletion can neither obtain contractual effect nor block subsequent republication; administrative reporting and civil litigation are the paths that leave a record of the handling.

Original text: https://www.cac.gov.cn/2026-08/04/c_1787588879723730.htm

II. The Cybersecurity Review Office Launches a Cybersecurity Review of Products Sold in China by Palo Alto Networks

On August 6, 2026, the Cybersecurity Review Office issued the Announcement on Launching a Cybersecurity Review of Products Sold in China by Palo Alto Networks. The body of the announcement is a single sentence, which can be read in three layers. The normative basis is two laws, the National Security Law of the People’s Republic of China and the Cybersecurity Law of the People’s Republic of China; the procedural basis is the Cybersecurity Review Measures; and the object of review is the products sold in China by Palo Alto Networks. The reasons given in the announcement are to safeguard the stable operation of critical information infrastructure, prevent risks and hidden dangers, and maintain national security. Beyond this, the announcement discloses no detail whatsoever on the scope of the review and makes no conclusory judgment of any kind.

This announcement enters this month’s reputation governance horizon not because of the review itself, but because of the way the review announcement, as an information source, is disseminated secondhand. Launching a review and reaching a conclusion are two different things, and the announcement makes no characterisation at all; yet along the dissemination chain, official documents of this kind are very easily processed into headlines carrying a conclusory implication. Article 9 of the SPC Online Infringement Provisions has a rule directly on point: where a network user or network service provider publishes information based on information sources such as documents produced by state organs in the exercise of their functions and acts of authority publicly carried out, and the information published is inconsistent with the content of the information source, or causes misunderstanding by adding insulting content, defamatory information, or improper headlines, or by adding or deleting information, adjusting the structure, or changing the order, thereby infringing the personal rights and interests of others, and the infringed party requests that it bear tort liability, the people’s court shall support the request; the same applies where the information source has been publicly corrected and the publisher refuses to make the correction or continues to publish the pre-correction information.

For multinational enterprises operating in China, an operable criterion can be drawn from this rule: when facing negative dissemination that takes an official announcement or regulatory document as its hook, whether enforcement can stand depends on whether the reporting goes beyond the document itself — whether it alters the scope of what the document states, whether it uses a headline to turn “launching a review” into a conclusion, and whether it continues to disseminate an old version after the document has been corrected. The mere fact that the reporting cites the document proves nothing. This follows the same logic as this month’s main theme: what makes an enforcement action stand is the granularity of evidence after item-by-item comparison, not a blanket view that “whatever is unfavourable to us is infringement.”

Original text: https://www.cac.gov.cn/2026-08/06/c_1787764332950791.htm

Reputation Enforcement Cases This Month

I. Li Auto v. “Dianche Xiaofei” Commercial Disparagement Case: Judicial Determination and Administrative Action Converge on the Same Account

Factual Background: According to an article by Chang’anjie Zhishi (Chang’an Avenue Insider) published on the Beijing Daily app, the car-review self-media account “Dianche Xiaofei” had long published untrue information and misleading content concerning Li Auto and related personnel; according to the same report, the account made leading insinuations about the R&D origins of Li Auto’s intelligent driving system and produced the short-video series “Li Auto Oddball Talk” (《理想奇葩说》) to publish negative comments (Source: Beijing Daily). Among the typical cases of the “Qinglang · Optimising the Online Business Environment — Rectifying Malicious Hyping of Enterprise-related Information” special campaign publicly exposed by the CAC on August 4, 2026, this account was listed first; the notice found that it had published misleading information, disparaged the product reputation and business reputation of an automobile enterprise, and profited from the resulting marketing and traffic diversion (Source: Cyberspace Administration of China).

Outcome or Current Status: According to a Beijing Daily app report, Li Auto’s legal department announced on Weibo on July 22, 2026 that, after first instance and second instance, the court had held that the conduct of “Dianche Xiaofei” constituted infringement, and the judgment expressly stated that its conduct constituted commercial disparagement (Source: Beijing Daily). According to the CAC notice referred to above, the three accounts involved have been closed in accordance with laws and platform agreements (Source: Cyberspace Administration of China). As of August 21, 2026, no public judgment in the case could be found, and neither the case number nor the full text of the judgment has been disclosed; this article does not restate the content of the judgment beyond public reporting.

Legal Issues: First, the line between commercial disparagement and legitimate reviews. Article 12 of the Anti-Unfair Competition Law (2025 Revision) regulates the fabrication or dissemination of false or misleading information that damages the business reputation or product reputation of other business operators; Article 8 of the SPC Online Infringement Provisions supports a business entity’s infringement claim from the angle of damage to public trust in the business entity and the lowering of the social evaluation of its products or services. According to a DoNews Auto report published on QQ.com, the court focused on whether the content was supported by objective data and uniform evaluation standards, not on whether the evaluation itself was negative (Source: DoNews Auto (QQ.com)) — and this is precisely the watershed between legitimate reviews and disparagement. Second, the effect of disclaimer language. According to the same DoNews Auto report, the content at issue carried disclaimer wording along the lines of “authenticity not guaranteed,” but the court did not exempt the account from liability on that basis (Source: DoNews Auto (QQ.com)); this is consistent with the analytical structure of Article 1025 of the Civil Code: the public-opinion-oversight exemption presupposes that facts were not fabricated or distorted and that a reasonable duty of verification was fulfilled, and a one-line disclaimer cannot substitute for verification. Third, the relationship between administrative action and civil remedies. Account closure addresses the continued spread of information; the business reputation losses an enterprise has already suffered must still be pursued separately through civil channels, and the two lines run in parallel without conflict.

Counsel’s Observations: The value of this case lies in placing two paths on the same account. Civil litigation delivers confirmation of rights and obligations — a long cycle but a stable conclusion; administrative reporting delivers speed and level of action — the notice makes clear that the leads came from letters sent by relevant departments and local cyberspace administrations, showing that reporting is the front end that triggers action and that the granularity of action can reach the level of the account operator. For enterprises, this is not an either-or choice: continuing infringement during litigation can be stopped through reporting and platform complaints, while determinations formed in judicial proceedings can be submitted together as evidentiary material in subsequent reports, with their acceptance and the resulting action determined by the competent authority in the exercise of its functions. By contrast, the “Zhu Zhiliao” incident of the same month followed a single line of platform complaints, with the granularity of action stopping at individual items of content and the complaining party having to bear the screening costs and online reputation costs itself — the difference between the two approaches seen in the same month is precisely the question this month’s main theme sets out to answer.

Implications for Enterprises: When facing smears at scale, administrative reporting is a path with relatively low cost and a relatively high level of action, and is worth evaluating alongside platform complaints and civil litigation, rather than defaulting to filing complaints item by item on platforms oneself. Material preparation for the three paths overlaps heavily — publication times, account homepages, original content, dissemination data, and leads on profit links are all shared, so a single round of preservation can support several paths. Two further points deserve attention: first, on the pace of enforcement, in this case the judicial conclusion came before the administrative notice, and after obtaining a judicial determination an enterprise still needs to bring it to regulatory channels in parallel, otherwise action may stop at the level of the individual case; second, on external messaging, the CAC notice did not name the enterprise involved, and an enterprise should not, in its own communications, use the notice to insinuate anything about competitors or unilaterally expand the scope of the determination.

Key Online Reputation Incident Review This Month

Incident Overview

According to a report by The Paper on August 4, 2026 and the Statement on Recent Online Information Relating to the “Zhu Zhiliao” published the same day by the official account “HIMA Spokesperson,” the origin of this incident can be traced back to the phrase “the best SUV under RMB 10 million” at the AITO M9 launch event at the end of 2023, and the gasps of the audience at the venue; the clip later became an internet meme. In late July 2026, the traditional bamboo toy “Zhu Zhiliao” (bamboo cicada), which makes a sound highly similar to that clip when its cotton cord is pulled, was used by short-video creators for derivative creation. According to an online reputation analysis published by Civiw, some creators’ approach subsequently escalated to using artificial intelligence to imitate the voice of the executive concerned, compositing his likeness into images of the toy, and registering insulting domain names to build interactive web pages. According to Civiw’s online reputation analysis, in late July the HIMA legal team filed batch complaints with multiple platforms on the grounds of infringement of reputation rights and damage to corporate business reputation; The Paper’s account was that Huawei then filed complaints against some of the videos, and the videos concerned were taken down.

Online Reputation Flashpoints

After the complaints were filed, claims quickly appeared online such as “‘Zhu Zhiliao’ toys taken down across the entire internet” and “ordinary nostalgic videos taken down by association, with account avatars greyed out and nicknames reset.” According to Civiw’s monitoring parameters, in the window from 00:00 on July 30 to 15:00 on August 5, 2026, the online reputation heat index for the topic was approximately 250,357, peaking at 22:00 on August 4; over that period, the distribution was approximately 48.1% social networks, approximately 28.7% news apps, and approximately 21.5% short video, and by sentiment 52.8% negative, 37.5% neutral, and 9.7% positive. According to The Paper, after the complaints even more mocking content of the same kind appeared online, and the reporter could still find large numbers of related videos by searching “Zhu Zhiliao” on Douyin and Xiaohongshu. Civiw summarised the process as the complaints themselves amplifying the heat in reverse: sales of related products on e-commerce platforms rose, and short videos commonly used “word is this one will get taken down” as a traffic-drawing hook.

Legal Risk Points

The first is the risk in screening complaint targets. Under Article 1195, Paragraph 1 of the Civil Code, where a rights holder notifies a network service provider to take necessary measures such as deletion, blocking, and link disconnection, the notice shall include preliminary evidence of infringement and the true identity information of the rights holder; Paragraph 3 of the same article further provides that a rights holder who causes damage to a network user or network service provider through an erroneous notice shall bear tort liability. The larger the scale of a batch complaint, the more easily the granularity of item-by-item evidence is diluted, and the greater the exposure to erroneous notices.

The second is the risk in choosing the basis of rights. Imitated voices and uglified likenesses targeting an executive personally, and damage to business reputation targeting the enterprise, are two different paths of claim. The basis for the former is Article 1019 of the Civil Code — no organisation or individual may infringe another’s portrait rights by uglifying, defacing, or forging by means of information technology, or by similar means — and Article 1023, Paragraph 2 — the protection of a natural person’s voice shall be governed, mutatis mutandis, by the relevant provisions on the protection of portrait rights; the basis for the latter is Article 12 of the Anti-Unfair Competition Law (2025 Revision) and Article 8 of the SPC Online Infringement Provisions. The two paths differ in rights holder, focus of proof, and form of remedy; asserting them together can easily lead, in a given case, to a mismatch between the rights holder and the object infringed.

The third is the risk in determining the boundaries of speech. According to The Paper, Xia Hailong, a lawyer at Shanghai Shenlun Law Firm, considers that netizens buying and using the toys or filming short videos to share them is normal conduct, and that even using this to express a view obliquely is a legitimate right; Zhai Wei, an associate professor at East China University of Political Science and Law, considers that the short videos contain elements of mockery of a corporate executive, that in anti-unfair competition legal practice mockery of an executive can be regarded as damage to the business reputation of the enterprise itself, and that from a purely legal standpoint the conduct does indeed potentially constitute infringement — but he also points out that leading enterprises and their executives, as public figures, ought to show greater acceptance and tolerance of mockery and even criticism from the public. The disagreement between the two interviewed experts itself shows that the boundary between mockery and infringement is not self-evident in individual cases; this is also why Article 1025 of the Civil Code separately lists “fabricating or distorting facts” and “using insulting language or similar means to disparage another’s reputation” as exceptions.

Response Assessment

According to the statement referred to above, at 22:00 on August 4 the enterprise disclosed three sets of verifiable figures: as of that evening, the total volume of related online information monitored was more than 56,000 items, of which complaints were actually filed against 171, and platforms reported taking down 144; the complaints targeted specific infringing content rather than the “Zhu Zhiliao” toy itself; no complaint about the product had been filed with any e-commerce platform, nor had any e-commerce platform ever been asked to take it down or stop selling it. The same statement also denied, point by point, claims such as “taken down across the entire internet,” “sales banned,” and “ordinary videos taken down by association, with account avatars greyed out and nicknames reset,” and stated that it respects normal online expression and good-faith criticism. From the standpoint of verifiability, disclosing the total monitored volume, the actual number of complaints, and the number of platform takedowns side by side is a form of statement that can be tested externally — it turns “whether this was one-size-fits-all” from a dispute over attitude into a dispute over proportions. According to Civiw’s monitoring parameters, the online reputation peak occurred at the same moment the statement was published; as of August 20, 2026, its published monitoring window ended at 15:00 on August 5, and no data on the same parameters is yet available for verifying the subsequent trend.

It should be noted that this section discusses only the verifiability and dissemination effects of the response and does not evaluate whether the enterprise’s enforcement conduct was lawful; whether the content concerned constitutes infringement is a matter of case-specific determination and falls outside the scope of this article’s analysis.

Implications for Enterprises

The lesson this month’s incident leaves enterprises sits at an earlier stage: how an enforcement action is understood by outsiders. Whether to enforce at all is, by contrast, not the point of contention. If a batch of complaints that is legally beyond reproach presents itself externally only as content disappearing, the public will fill in the motive with whatever fragments are at hand, and the narrative so filled in is often harder to deal with than the original dispute. The two trends noted above — heat and sales rising rather than falling, and similar content increasing rather than decreasing — trace back to the same mechanism: the act of deletion itself became new dissemination material, through which the motive narrative filled in by the public continued to spread.

The second layer of lessons concerns granularity. Article 1195 of the Civil Code makes preliminary evidence of infringement an element of the notice, while leaving damage caused by an erroneous notice to be borne by the rights holder; the practical meaning of this pairing is that batching is only a mode of execution, and liability is always item by item. The strength of an enterprise’s content governance capability shows not in how many items it can submit at once, but in whether it can state clearly, for each item, which norm it falls under and which piece of evidence it relies on; doing this up front is both a defence against liability for erroneous notices and the only material that will stand up when an external explanation is later needed.

The third layer of lessons concerns the two sides of the same capability. An enterprise may be the party enforcing rights, or the party held accountable; if it demands externally that platforms act on evidence, it must be able to show internally that the publication chains of its own accounts and authorised accounts carry review records to an equivalent standard — another widely reported controversy over marketing copy this month shows that handing operations to a third party does not change where the obligations belong. Finally, timing: the value of a clarification diminishes along the emotional curve; the later a statement is published, the more easily it is read as a passive response, and every factual assertion in it may be submitted as documentary evidence in subsequent proceedings. Moving the preparation of messaging forward to before the action is launched is, in incidents of this kind, the lowest-cost arrangement and also the one most easily overlooked.

In-Depth Theme This Month

Theme: The Compliance Boundaries of Enterprise Online Rights Enforcement and Online Reputation Foresight — From Screening Complaint Targets to Preventing Backlash

When enterprises take the initiative against infringing online content, the question has now come down to how to do so without getting themselves caught up in it. According to HIMA’s Statement on Recent Online Information Relating to the “Zhu Zhiliao” of August 4, 2026, the targets of action it listed fell into four categories: out-of-context editing of launch-event audio, AI forgery imitating an executive’s voice, AI photo editing that uglifies an executive’s likeness and the making of insulting memes, and the registration of insulting domain names to carry out personal defamation; Civiw’s monitoring parameters show that from July 30 to August 5 the online reputation heat index for the incident was approximately 250,357, with negative sentiment accounting for 52.8%.

I. Layered Screening of Enforcement Targets

Screening begins with one question: which prohibitive norm does this content fall under? Making the enterprise uncomfortable is not in itself a reason for selection; content that cannot be placed under a provision does not go on the enforcement list.

AI uglification or alteration of an executive’s image falls under Article 1019 of the Civil Code — no one may infringe another’s portrait rights by uglifying, defacing, or forging by means of information technology, or by similar means; the prohibition in this article does not presuppose use for profit, and the assessment does not depend on whether the maker monetised the content. AI-synthesised imitation of a voice falls under Article 1023, Paragraph 2, under which the protection of a natural person’s voice is governed, mutatis mutandis, by the relevant provisions on the protection of portrait rights, with identifiability as the threshold: the public must be able to identify a specific natural person from timbre, intonation, and pronunciation style. This threshold draws the line between two forms: for the forms discussed in this article, an aural overlap between a toy’s sound effect and an audience’s gasps is generally insufficient to produce identification, whereas synthesised speech that makes the public believe someone said something they never said is another matter; how an individual case is determined still depends on the specific content and evidence.

Editing works the same way. The act of editing itself is assessed under Item (2), Paragraph 1 of Article 24 of the Copyright Law of the People’s Republic of China, namely that a published work of another may be appropriately quoted in order to introduce or comment on a work or to explain a point; the line lies not in whether something was edited, but in whether a factual meaning absent from the original material was manufactured, and only verifiable untrue statements of fact and insulting expression come within the reach of reputation rights under Article 1024 of the Civil Code. Insulting domain names are the category in which form and substance are most clearly separated: registration produces only a formal credential, and where the readable meaning of the domain name is itself a disparagement of a specific natural person, it still falls under reputation rights and personal dignity.

The other side of the boundary is held by Article 1025: a person who carries out news reporting, public opinion oversight, or similar conduct for the public interest and thereby affects another’s reputation does not bear civil liability, except in three situations — fabricating or distorting facts; failing to fulfil a reasonable duty of verification with respect to seriously untrue content provided by others; and using insulting language or similar means to disparage another’s reputation. These three items are the most practical self-check list before a complaint — if content falls under none of them, the complaint is legally weak.

II. The Personality Rights Path and the Business Reputation Path Are Two Separate Lines

Infringement of personality rights targeting an executive personally and damage to business reputation targeting the enterprise belong to two separate paths of claim, and are the most easily flattened in batch handling. The former proceeds under the Part on Personality Rights, where the rights holder is the natural person himself or herself and the enterprise cannot automatically act in their place. The latter has two anchors: commercial disparagement regulated by Article 12 of the Anti-Unfair Competition Law (2025 Revision), namely fabricating or disseminating false or misleading information that damages a competitor’s business reputation or product reputation; and Article 8 of the SPC Online Infringement Provisions — where defamation, disparagement, or similar means are used to damage public trust in a business entity or lower the social evaluation of its products or services, the people’s court supports the business entity’s claim for tort liability. The latter does not presuppose a competitive relationship and has broader coverage.

The distinction has three layers of consequences: the object of proof differs — personality rights turn on identifiability and disparagement, business reputation on the lowering of social evaluation and the impairment of trust; the rights holder differs — for speech targeting an individual to be converted into a claim by the enterprise, a contextual link must be separately proved; and the remedy differs — compensation for mental distress can be claimed only by a natural person. The mechanisms established by the Provisions on Governance of Cyber Violence Information are directed mainly at natural-person users and should not be applied directly to the enterprise business reputation line. Packing the two types of content into the same batch amounts to asking a platform to adjudicate two kinds of legal relationship in a single notice.

III. The Double-edged Nature of the Notification-and-Necessary-Measures Mechanism

Article 1194 of the Civil Code establishes the general rule that one who uses the internet to infringe another’s civil rights and interests shall bear tort liability; Article 1195 provides that a rights holder is entitled to notify a network service provider to take necessary measures such as deletion, blocking, and link disconnection, that the notice shall include preliminary evidence of infringement and the true identity information of the rights holder, and that a platform that fails to take necessary measures in a timely manner bears joint and several liability with the network user for the extended part of the damage.

The double edge lies in the last sentence of Article 1195 of the Civil Code: a rights holder who causes damage to a network user or network service provider through an erroneous notice shall bear tort liability. The companion provision is Article 5 of the SPC Online Infringement Provisions — where a network user whose information has been deleted asserts liability against the platform and the platform raises the defence of having received a notice that complies with the rules, the people’s court shall support the defence. When the notice is valid, the platform’s liability is cut off and the risk falls back on the rights holder.

On this basis, the point at which the risk lands can be stated more specifically: the legal risk of batch complaints lies not in “batching” as a mode of execution, but in the granularity of the evidence for each notice. The preliminary evidence requirement of Article 1195 applies to each notice, not to a batch; a list generated by batch scraping is merely a collection of mutually independent notices, and any one item that fails constitutes a risk exposure on its own.

IV. The Order of Selection of Enforcement Tools and Their Triggering Scenarios

Platform complaints suit content that clearly falls under a prohibitive norm, where the actors are unknown or highly dispersed and dissemination needs to be cut off quickly; they have the lowest cost but the poorest explainability, as outsiders see only that content has disappeared. Administrative reporting suits content that has the character of illegal or harmful information or that concerns the order of enterprise-related information; the administrative authority makes the judgment, and the enterprise need not characterise the other party’s conduct itself. The threshold for a personality rights injunction under Article 997 of the Civil Code is much higher: there must be evidence that the actor is carrying out or is about to carry out an illegal act infringing personality rights, and that failure to stop it in time would cause irreparable harm to lawful rights and interests — corresponding to infringement that is continuous, high-frequency, and expected to continue; a one-off single item of content will mostly fail the urgency element. Civil litigation suits cases requiring a confirmatory conclusion or damages, at the cost of a long cycle and the risk that the litigation itself becomes a news source. Criminal reporting targets the fabrication and dissemination of false facts causing heavy losses or involving other serious circumstances, and whether the threshold for prosecution is met is determined by the judicial authorities; for public insult and defamation by fabricated facts targeting an executive personally, Article 50 of the Public Security Administration Punishments Law of the People’s Republic of China (2025 Revision) establishes a separate administrative penalty path, with natural persons as the protected party. On ordering, there are two practical considerations: place recoverable actions before unrecoverable ones, and place paths in which a third party makes the judgment before paths in which the enterprise characterises the conduct itself. This is a trade-off based on cost and reversibility, not a statutory order.

V. Preventing Backlash: Designing the Explainability of Enforcement Actions

Explainability has three handles that can be designed in advance. The first is stating the parameters separately: this month’s statement set out at once that 56,000 items had been monitored cumulatively across the internet, that complaints had been filed against only 171, and that platforms had taken down 144 — giving the denominator and the numerator together, which amounts to replacing the adjective “batch” with a verifiable question of proportion. The second is the affirmative statement of exclusions: spelling out what is not being targeted usually constrains the narrative more strongly than repeatedly emphasising what is being targeted, and the statement denied three circulating claims point by point. The third is timing: monitoring parameters show that the incident’s online reputation peaked at 22:00 on August 4, and the statement was also published at 22:00 that day; when a clarification appears at the same time as the peak, the text of the clarification itself also becomes dissemination material. Every sentence in a statement should withstand being excerpted on its own — an external statement is documentary evidence and may be submitted by the other side in subsequent proceedings, and once absolute expressions such as “never” or “all” are disproved in an individual case, the cost exceeds the persuasive force they bought.

VI. Joint Reputational Risk Along the Outsourcing Chain

When a brand-authorised account is run by a third party, the public sees the brand, but legal liability is allocated among the advertiser, the advertising operator, and the advertising publisher. According to an article by the People’s Daily Online Public Opinion Data Center, Shanghai Baishide Electronics Co., Ltd. (上海百事得电子有限公司), a distributor of Logitech, was fined RMB 200,000 by the Shanghai Jing’an District Market Supervision Administration for posting on an official Douyin account a price-cut video captioned “Once I cut my price, won’t you come running like a dog”; according to the same article, on the night of the controversy the brand’s Weibo account apologised, stating that the non-compliant content had been posted without authorisation by an employee of the authorised company who skipped the review process, and that apology in turn triggered secondary online reputation over content review and the control of outsourced operations. Item (7) of Article 9 of the Advertising Law of the People’s Republic of China prohibits advertisements from containing content that disrupts public order or violates good social morals, and Article 57 sets a fine of RMB 200,000 to 1,000,000 for advertisers who violate that article; marketing by disparaging consumers touches on the right of consumers to have their personal dignity respected, as confirmed by Article 14 of the Law of the People’s Republic of China on the Protection of Consumer Rights and Interests. External enforcement and internal governance draw on the same set of capabilities: if an enterprise asks platforms to handle its complaints at the granularity of evidence, it must be able to show that the publication chains of its own and authorised accounts carry review records of the same granularity — otherwise the persuasiveness of its enforcement claims will be undercut by the loosest link in its own chain.

Counsel’s Prescription

I. Legal Risks an Enterprise May Face When Launching Online Rights Enforcement, and Response Strategies

The legitimacy of an enterprise proactively clearing infringing online content is long established, but enforcement actions themselves also generate new legal risks, which do not dissolve merely because the purpose of enforcement is legitimate. There are three main categories: the risk of compensation for an erroneous notice when a complaint fails; the risk that a mismatch between the path of claim and the form of content causes the claim to fail; and the risk that the enforcement means itself is evaluated negatively or even found to constitute commercial disparagement. The common trigger for all three is insufficient granularity in complaint decision-making.

When the double-edged nature of the notification-and-necessary-measures mechanism discussed above is translated into operations, the first and third categories of risk share the same judicial review path. In the commercial disparagement dispute between a certain network company and a certain biotech company (Typical Case No. 6 among the Supreme People’s Court’s typical cases on the protection of the personality rights of private enterprises and private entrepreneurs), the triggering scenario was that a manufacturer, dissatisfied that another store on the same platform sold the same product at a lower price and having repeatedly demanded a price adjustment without success, filed an intellectual property infringement complaint on the grounds that the rights holder had never produced or authorised the production of that model, causing the other party’s products to be removed and its store to lose points; the court examined subjective intent along four dimensions — the cause of the complaint, the manner and grounds of the complaint, the purpose of the complaint, and whether remedial measures were taken after the complaint — and found that the conduct constituted commercial disparagement. Of these four dimensions, the first three can be fixed before a complaint is filed: whether the motive is mixed with commercial considerations unrelated to the determination of infringement, whether the grounds correspond one-to-one with the evidence attached, and whether the purpose goes beyond stopping the infringement itself; the fourth depends on the handling after the complaint. On this basis, we recommend turning complaint decisions into an item-by-item ledger with two kinds of fields. Those mandatory before the complaint correspond one-to-one to the first three dimensions: the reason for initiating this item, together with confirmation that it is unrelated to commercial demands such as pricing or competition (corresponding to motive); the provision relied on, the corresponding evidence number, and the time of the determination (corresponding to grounds and evidence); and confirmation that the scope of the complaint request for this item is limited to that item of content itself (corresponding to purpose). Those filled in after the complaint are the outcome of handling, and the record of any application to the platform for restoration upon discovering that an item lacked a basis. The latter both corresponds to the item “whether remedial measures were taken after the complaint” among the four dimensions and serves as material that can be submitted when later proving good faith.

Adjacent to this is the mismatch of claim paths, which determines whether enforcement can succeed at all. In the online tort liability dispute between a certain technology company and Li [redacted] (Typical Case No. 4 among the Supreme People’s Court’s typical cases on judicial protection of enterprises’ reputation rights), the triggering scenario was that a self-media practitioner published disparaging content targeting a technology company and its legal representative on multiple accounts he operated, and the remarks evaluating the legal representative were all published in a context evaluating the company’s business conduct; taking into account the surrounding context of the remarks and the understanding of the general public to determine their target, the court held that where the founder’s reputation is closely linked to the enterprise’s reputation, the enterprise may assert rights over the remarks at issue, including those targeting the founder. This rule cuts both ways: for an enterprise to assert its own rights over remarks targeting an executive, it must be able to prove a contextual link; if the disputed content concerns only the executive’s personal life and has nothing to do with the business, a complaint in the enterprise’s own name lacks a basis, and the claim should be pursued by the individual through the personality rights path. We recommend completing a target determination when the enforcement project is initiated, and retaining the basis for that determination together with the original carrier of the content; domain names, accounts, and trademark registrations whose readable meaning disparages an executive’s name should be gathered and monitored separately, as their handling path differs from that of ordinary negative content.

How much compensation can be obtained is ultimately determined by evidence. According to a Benliu News report of March 18, 2026, the Henan High People’s Court disclosed that day the trial of the reputation rights dispute brought by Xuchang Pangdonglai Commerce and Trade Group Co., Ltd. (许昌胖东来商贸集团有限公司) against Duan [redacted]: in January 2025, Duan [redacted], a blogger with a million followers, published a disparaging video at a time when the merchant had already responded to a complaint and the cause of an allergic reaction had not yet been determined; the video was viewed 12.174 million times and received 114,000 likes and 42,000 comments; the enterprise successively sent the products concerned to three quality inspection institutions for testing and took the relevant category of products off the shelves. The court held that Duan [redacted] had failed to fulfil a basic duty of verification and that his subjective fault was evident, constituting an infringement of the enterprise’s reputation rights; taking into account factors such as brand influence, the scope of dissemination, the actual losses the enterprise incurred from removing products and sending them for testing, the degree of fault, and his attitude in admitting wrongdoing, the court ordered him to publish on his Douyin account a written apology approved by the court and keep it undeleted for 30 days, and to pay compensation for economic losses totalling RMB 400,000 (Source: Benliu News); according to the same report, neither party appealed after the judgment was pronounced (Source: Benliu News). This case lays out the basis on which compensation is calculated: the ceiling of the amount is determined by the evidence already gathered, not by the amount claimed. Payment vouchers for testing, notarisation, preservation, and legal services, together with platform-side data on views, comments, and reposts, should be preserved item by item during the period in which the incident occurs — doing so after the fact costs far more, and some data can no longer be traced back by then.

II. Four Things That Can Be Started Immediately

Turning to arrangements that can be started immediately, the first is to build the item-by-item determination form. For each item of content proposed for complaint, record at least five items: the form of the content; whether the target is the enterprise or an executive personally; the specific provision to be asserted; the corresponding evidence number; and whether it falls within one of the three exceptions listed in Article 1025 of the Civil Code. Content that cannot be placed under a provision should not be included in the current batch. This is not an additional procedural burden — Article 1195 requires a notice to include preliminary evidence of infringement, which is itself a requirement on each notice.

The second is to place unrecoverable actions later. Platform complaints, letters, and demands for rectification are recoverable actions; lawsuits, public characterisation, and criminal reports are unrecoverable actions. Clear escalation conditions should be set between the two, rather than advancing on emotion. This month’s incident points to an easily underestimated transmission: once an enforcement action is restated by an online narrative, it may in turn amplify the dissemination of the original content; preparing external messaging in parallel before launching actions at scale is usually better than waiting for online reputation pressure to force a catch-up.

The third is to review external texts for excerptability — as noted above, an external statement is documentary evidence that may be submitted by the other side in subsequent proceedings, and every sentence must withstand being excerpted on its own — concretely in three aspects of drafting: add qualifiers to statements of fact, such as “based on verification to date” or “according to information available as of now”; the other two — the affirmative statement of what is not being targeted, and presenting monitoring parameters and complaint parameters separately — have been explained in the section above on designing explainability, and need only be applied to the text. All external voices, including outside agents, will be taken by the public by default as the enterprise’s own position, so we recommend agreeing on external communication messaging and the boundaries of expression at the time the engagement is established.

The fourth is to align the publication chains of own accounts and authorised accounts to the same standard — the obligations under the Advertising Law and the Consumer Rights and Interests Protection Law discussed above do not transfer merely because operations are outsourced. We recommend specifying in authorisation and outsourced-operation agreements the content review stage, record-keeping requirements, and the allocation of liability after a violation, and bringing core partners within the scope of routine reputation monitoring.

Governance Toolbox This Month

This Month’s Toolbox: Compliance Self-check Checklist for Enterprise Online Rights Enforcement Actions

This checklist is for item-by-item screening before launching enforcement actions such as batch complaints, lawyer’s letters, reports, or lawsuits. Usage instructions: ask yourself each row in turn; if the answer to any row is unsettling, first adjust according to the action in the last column, then decide whether to proceed. For rows that list a normative basis, the criteria come from the provisions themselves; the last row has no normative basis and is a recommendation at the level of online reputation contingency planning, as separately indicated.

Dimension	Question to Ask	Normative Basis	Action If the Answer Is Unsettling
Ownership of the basis of rights	Is what has been infringed the personality rights of an executive personally, or the business reputation and product reputation of the enterprise? Have the two been mixed together and asserted in the same set of materials?	Individual side: Articles 1019 and 1023, Paragraph 2 of the Civil Code; enterprise side: Article 12 of the Anti-Unfair Competition Law (2025 Revision), Article 8 of the SPC Online Infringement Provisions	Split into two sets of materials asserted separately, matching each item to its rights holder and the object infringed
Evidentiary granularity of the notice	Does each item complained of independently carry preliminary evidence of infringement? Does the notice state the true identity information of the rights holder?	Article 1195, Paragraph 1 of the Civil Code	Withdraw items with insufficient evidence from the complaint list rather than submitting them as a package
Exposure to erroneous notices	Does the complaint list include normal comments, criticism, or derivative works that do not constitute infringement? If they are hit by mistake, how much damage could result?	Article 1195, Paragraph 3 and Article 1025 of the Civil Code; Item (2), Paragraph 1 of Article 24 of the Copyright Law	For borderline items, switch to preserving evidence and observing first, without complaining for now
Official documents as information source	Does the negative content take as its information source a document produced by a state organ in the exercise of its functions? Does the reporting go beyond the document itself?	Article 9 of the SPC Online Infringement Provisions	Compare the original document and the reporting text sentence by sentence, and assert claims only over the parts that go beyond it
Matching the handling path	Platform complaint, administrative report, personality rights injunction, civil litigation, criminal report — which one truly matches the factual pattern of this case?	Article 997 of the Civil Code; Article 28 of the Anti-Unfair Competition Law (2025 Revision); Item (2) of Article 50 of the Public Security Administration Punishments Law (2025 Revision)	Order from low to high irreversibility, placing the least reversible action last
Risk of paid deletion	Is the other party demanding money in the name of deleting posts or eliminating impact? Is anyone internally advocating paying to make it go away?	Article 10 of the SPC Online Infringement Provisions	Such agreements have no effect; switch to reporting and litigation and keep a complete record of the handling
Rights and responsibilities along the outsourcing chain	When a brand-authorised account is run by a third party, how are rights and responsibilities for content review and compliance accountability allocated? Who bears external responsibility once non-compliant content is published?	Item (7) of Article 9 and Article 57 of the Advertising Law; Article 14 of the Consumer Rights and Interests Protection Law	Write content red lines and an accountability mechanism into the authorisation agreement; do not substitute after-the-fact apologies for advance review
Explainability of the action	Once the enforcement action is publicly discussed, can its scope be explained with verifiable figures, such as the ratio among total monitored volume, actual complaints, and platform actions?	No direct normative basis; an online reputation contingency planning recommendation	Prepare messaging and data before launching batch actions, rather than catching up after online reputation builds

It should be noted that this checklist answers “whether it can be done, and how to do it more safely,” not “whether doing it will win.” Whether infringement is found in an individual case depends on the specific content, scope of dissemination, and state of the evidence, and falls within the authority of judicial or administrative bodies to determine; this checklist makes no prediction.

Disclaimer

This article is intended solely for risk analysis purposes and does not render any determination of illegality or infringement against any party; the merits of individual cases shall be determined by effective judicial decisions.