Influencers under attack online: pile-on harassment, deepfakes and fake profiles

For a content creator, reputation is not a matter of sensitivity: it is the means of production. It carries the brand deals, the monetisation, the audience. A coordinated campaign, a fabricated video or a fake account therefore translates directly into lost revenue — and French criminal law has, over the past few years, been sharpened against precisely these three forms of attack.

Pile-on harassment: each participant answers, even for a single message

The argument one hears most often, when a creator is buried under a wave of hostile messages, is also the weakest: “I only posted one comment.” Since the Law of 3 August 2018, it protects no one.

Article 222-33-2-2 of the French Criminal Code punishes harassment and expressly covers two pile-on scenarios: where the statements are imposed on the same victim by several people acting in concert, even though none of them acted repeatedly; and where they are imposed successively by several people who, even without any concerted plan, know that their statements form part of a repetition. In other words: whoever adds their message to a wave they know is already under way answers for harassment, individually and in their own right.

Two practical consequences. First, the strategy is not to sue a thousand accounts but to identify the instigators and the most clearly established participants — those whose contribution shows knowledge of the ongoing wave. Second, the “deterioration of living conditions” the provision requires must be documented: describing the pile-on is not enough, its effects must be established, medically and personally.

Deepfakes and doctored content: what the SREN Law changed

The misuse of image and voice through artificial intelligence has become a front-line risk for creators — fictitious product placement, words never spoken, fabricated intimate content. The SREN Law of 21 May 2024 modernised article 226-8 of the Criminal Code to address it expressly.

The decisive element is labelling: the offence presupposes that it is not obvious the content is a montage or generated, and that no express mention says so. A clearly identified parody falls outside the provision; a video faithfully imitating a genuine statement falls squarely within it. The qualification therefore turns on the appearance of authenticity — and the online aggravation, unusually severe compared with the base offence, reflects the legislature’s intent to strike at viral distribution.

Fake profiles and identity theft

An account borrowing a name, a photograph and a tone to address a community falls under article 226-4-1 of the Criminal Code: usurping a third party’s identity, or using data identifying them, in order to disturb their peace or damage their honour, is punishable by one year’s imprisonment and a €15,000 fine — with identical penalties where the acts are committed online. Unlike the deepfake provision, the legislature did not add a digital aggravation here: the offence was designed for the digital world from the outset.

This criminal route usefully combines with a civil claim. For a fake profile, French courts compensate both the infringement of the right to privacy and the infringement of image rights — two distinct heads of damage, which accumulate. It is often the fastest route to obtain removal of the account together with compensation.

The takedown video: never act without a full transcript

The format has become a genre of its own: a long, documented video putting a creator on trial before hundreds of thousands of viewers. The temptation is to rush to court with selected excerpts. That is the mistake to avoid.

The Court of Cassation requires an assessment of the statements as a whole: in order to determine their true meaning, “the offending statements must not be taken in isolation but interpreted in relation to each other” (Cass. 1st Civ., 11 May 2022, no. 21-12.063). A judge presented with excerpts cannot characterise them: the context, the tone and the factual basis relied on are missing. The discipline is therefore invariable — a bailiff’s report covering the video, its views and its comments, then a full transcript, before any cease-and-desist letter.

One final point, in the interest of candour: French case law specific to content creators remains sparse. Published decisions concern companies, journalists and elected officials; the litigation of the influence economy is only now taking shape. That does not weaken a victim’s position — the criminal provisions set out above are recent and precise, and press law applies in full — but it means reasoning by analogy, and taking particular care over proof of loss, in the absence of quantified precedents to rely on.

Quantifying the harm: the creator’s economic file

This is where a creator’s case genuinely differs. French courts do not compensate an alleged downturn: they require proof of the link between the attack and the loss. Margin, the share of the business affected, the certainty of the damage — the method is demanding, and it is prepared before proceedings begin.

Three reflexes, from the first hours: preserve the record of brand deals terminated, suspended or not renewed, together with the exchanges giving the reason; extract audience and revenue statistics from before and after the offending publication; document the costs incurred to contain the attack. Without those elements the harm remains moral — compensated, but on a scale unrelated to the actual loss.

On the choice of route — notice to the platform, delisting, civil interim relief or criminal proceedings — and on the three-month limitation period that governs press characterisations, see our comprehensive analysis: online reputation attacks: what remedies?

Frequently asked questions

Is a single comment within a wave of messages punishable?

Yes. Article 222-33-2-2 of the French Criminal Code expressly provides that the offence is committed where statements are imposed on the same victim by several people acting in concert, even though none of them acted repeatedly, or successively by people who know their statements form a repetition. A single contribution to an ongoing pile-on therefore engages its author.

Does a parody fall within the deepfake provision?

No, provided the fabricated nature is obvious or expressly stated: article 226-8 of the Criminal Code only punishes montages and algorithmically generated content where their nature is not manifest. It is the appearance of authenticity that triggers the offence, not the technical manipulation alone.

Can lost brand deals be compensated?

They can, but it must be proved. The court requires an established causal link between the offending content and the loss, assessed on objective evidence — terminated contracts, revenue data, margin, the share of the business affected. Screenshots and indignation are not enough: the economic file is built from day one.