How to remove content from social media: threats, cyberbullying or defamation?

A hostile social media campaign rarely consists of a single post. It may combine threats, repeated messages, photomontages, disclosure of an address and specific allegations. Removal requires each item to be isolated and matched to the correct legal regime.

A court may order removal of threatening or harassing content under Article 6-3 of the French LCEN. Defamation follows a different procedure: where the author is not a party to the proceedings, merely alleging that words are defamatory may not be enough to compel the platform to remove them.

Yes, where content causes harm and the requested measure is necessary, precise and proportionate. Article 6-3 of the French Digital Economy Act allows the president of the judicial court, under expedited proceedings on the merits, to order any person capable of contributing to take measures suitable for preventing or stopping that harm. A platform may therefore be targeted even though it did not write the post.

This power does not result in the indiscriminate deletion of everything a victim considers offensive. The court balances protection of the individual, privacy, personal data and freedom of expression. It examines each post, its context, repetition and the specific risk it creates.

The strongest application does not seek the deletion of an entire account as a matter of principle: it identifies each item, its legal characterisation, URL, date and resulting harm.

Characterisation determines the court, procedure, evidence and deadline. A death threat, call to hack an account or series of messages aimed at one person may fall under criminal law. An undisclosed photomontage may come within Article 226-8 of the French Criminal Code. Disclosure of an address may form part of cyberbullying, even if it does not alone establish the offence under Article 223-1-1.

Defamation follows another logic. It involves alleging a specific fact that harms a person's honour or reputation. Accusing someone of an offence or defined conduct is not legally equivalent to publishing a threat or degrading image.

The distinction creates no moral hierarchy between the harms. An allegation may be more destructive than a threat. But its unlawfulness depends on a debate about truth, good faith and context in which the author should normally be able to participate.

Explicit threats, incitement to commit an offence, unlawful photomontages, personal data and posts forming part of harassment may justify a targeted injunction. Repetition matters: one ambiguous message may form part of an organised campaign and reveal a risk when read with the others.

In its judgment of 31 August 2026, the Paris Judicial Court found that virulent and threatening stories, calls to hack accounts, degrading images and disclosure of an address formed part of unlawful harm capable of justifying removal. The platform removed the content during proceedings. The request therefore no longer required a ruling, but the platform was ordered to pay costs because litigation had been necessary.

Removal after proceedings begin does not make the case pointless: it may show that the action was justified and leave the platform liable for costs.

The French Act of 29 July 1881 strictly governs defamation claims. The author may rely on truth or good faith. Where proceedings involve only the victim and host, that adversarial debate cannot take place in the same conditions.

The Cour de cassation has therefore held that merely alleging defamation is insufficient to order a host to remove material. Exceptions may be considered where an earlier judgment has established that identical words are defamatory or the author cannot be identified and the requested measure remains proportionate.

In practice, all posts should not be grouped under one label. Threats and harassment may support action against the platform, while defamatory allegations will often require a claim against their author within the short three-month period.

Complete deletion is a particularly serious interference with freedom of expression. It is not justified where the account also contains lawful posts and only certain items have been identified as harmful. A broad, poorly evidenced request may be rejected when a precise list of URLs could have led to targeted removals.

To obtain account closure, the applicant must show that the account itself is devoted to attacks, that unlawful posts persist despite an earlier ruling and that no less restrictive measure can stop the harm. The proportion of lawful content, period examined and author's later conduct are decisive.

EU law prohibits a general obligation on intermediaries to monitor everything users publish. A targeted injunction may nevertheless be possible if it concerns precisely defined content, lasts for a limited period and can be implemented without a complex independent legal assessment by the platform.

The application should therefore define the equivalent content sought, the people concerned, identifying words, images or features and the duration. A vague request to prevent any future reputational harm transfers a judicial role to the platform that it cannot perform.

Preserve complete URLs, account identifiers, dates, comments, view counts and the sequence of publications. A screenshot alone may be disputed or become unusable if it shows neither address nor context. For a serious or evolving campaign, a judicial commissioner's report secures the evidence.

Then separate the material into categories. Threats, incitement, photomontages, personal data, insults and factual allegations may not use the same route. This mapping supports a reasoned platform notice and, if needed, proceedings that do not confuse the legal bases.

Act without delay. The standard limitation period for public defamation is three months. Material may disappear, accounts may change names and some identification data are retained only for a limited time.

An effective notice does not merely say that content is false or harmful. It gives the exact URL, quotes the disputed passage, explains its characterisation, identifies the person affected and provides evidence making the unlawfulness understandable without disproportionate investigation. The Digital Services Act notice-and-action mechanism complements LCEN remedies.

If the platform refuses or does not answer, preserve its moderation decision and all exchanges. They help prove knowledge of the content and steps taken before court proceedings. A formal notice based on an organised file is more useful than successive generic reports.

Expedited proceedings on the merits under Article 6-3 LCEN may seek measures suitable for stopping harm caused through an online service. Depending on the facts, a criminal complaint may also address threats, harassment or interference with a person's representation. Article 145 of the Code of Civil Procedure may obtain identification evidence before trial where there is a legitimate reason.

Public defamation is governed by the special formalities and calendar of the 1881 Act. The words must be characterised precisely and action brought within three months of first publication, subject to statutory interrupting steps. The wrong basis or a late writ can end the case.

INFLUXIO advises executives, artists, creators, agencies, brands and individuals facing social media campaigns. The firm begins with a post-by-post audit covering preservation, characterisation, urgency, identification of the author and choice of defendant.

Where appropriate, a legal notice and targeted formal demand are sent to the platform. If removal does not follow or danger requires an immediate response, the firm prepares the appropriate proceedings without confusing defamation with threats, image rights, personal data or cyberbullying.

The strategy also covers what follows removal: identifying the author, claiming compensation, preventing precisely equivalent republications and protecting online reputation over time. Contact the firm for an analysis of the content and the deadlines applicable to your situation.