Cabinet Kohen Avocats · Paris 16ᵉ
Ceuta Disinformation: What France Would Need to Prove Before Prosecuting a Cross-Border Organizer

Maître Hassan KOHEN intervient en urgence, du commissariat à la cour d'assises. Première analyse stratégique : 80 € TTC, réponse personnelle sous 24 heures.

100 % confidentiel · Secret professionnel · Sans engagement

Article généré par une intelligence artificielle, selon un processus conçu et contrôlé par le cabinet

Source : Cour de cassation – Base Open Data « Judilibre » & « Légifrance ».

Barreau de Paris Garde à vue, instruction, assises Fiche CNB avocat.fr
Maître Hassan KOHEN, avocat au Barreau de Paris
Maître Hassan KOHEN
Avocat au Barreau de Paris

Ceuta Disinformation: What France Would Need to Prove Before Prosecuting a Cross-Border Organizer

Parler à un avocat 06 89 11 34 45

The Ceuta border crisis has taken a new legal turn. On 31 August 2026, Spanish Prime Minister Pedro Sánchez said that networks linked to Russia, Israel and an international far-right movement had spread or amplified disinformation connected with the mass crossing into the Spanish enclave. The allegation remains a political attribution, not a criminal finding. The Associated Press has reported that many people travelled after seeing false claims that the border was open or that Spain would allow new arrivals to remain legally; other reporting puts the July 30–31 influx at roughly 72,000 people. Madrid has also rejected the idea that Moroccan state authorities organised the movement.

The urgent legal question is narrower than the political argument: what would France need to prove before treating an online campaign as a French criminal case involving a cross-border organiser? A false message is not automatically a crime. Investigators must identify a defined offence, link a particular person to a particular act, establish the required intent, and show why French jurisdiction and French evidence rules apply. A viral post can be morally troubling, strategically coordinated or commercially exploited without satisfying every element of complicity or direct provocation.

This analysis applies the French and European procedural perspective to the facts publicly reported about Ceuta. It does not identify Russia, Israel, Morocco, a platform or any named account as criminally responsible. It explains the proof that a Spanish or French investigation would have to preserve, the limits of the French criminal law, the use of European investigation orders, and the safeguards available to suspects, platforms and victims.

I. What would make the Ceuta online campaign a French criminal case?

A. Is a false border-opening message an offence on its own?

The first legal distinction is between falsity and criminality. A post stating that “the border is open” may be false, misleading, incomplete or based on a distorted reading of a court ruling. None of those descriptions, by itself, identifies a French offence. The prosecution must move from the message to a statutory provision whose elements fit the message, its context and its consequences. That discipline matters in Ceuta because the public debate has mixed at least four different propositions: that a border rule changed, that people could remain in Spain, that a route was safe, and that an outside actor deliberately engineered the movement. Each proposition requires separate evidence.

The constitutional starting point is freedom of expression. In Decision No. 2020-801 DC of 18 June 2020, the Constitutional Council examined online content restrictions and stated that “les atteintes portées à l’exercice de cette liberté doivent être nécessaires, adaptées et proportionnées à l’objectif poursuivi”. The sentence is not a licence to publish dangerous instructions. It is a warning that a criminal or administrative response must be tied to a real legal abuse and proportionate to the objective pursued. It also explains why a prosecutor should preserve the entire message, not only a headline or an edited screenshot.

The most direct French press-law route is a genuine call to commit an offence. Article 23 of the Law of 29 July 1881 expressly covers electronic communication and provides that persons may be punished when they “auront directement provoqué l’auteur ou les auteurs à commettre ladite action, si la provocation a été suivie d’effet”. Three conditions are immediately visible in that wording: the message must directly provoke, the target conduct must be an identified crime or offence, and the provocation must have produced the legally relevant result. Merely reporting that a route exists, repeating a rumour without an instruction, or criticising Spain’s border policy is not automatically that form of provocation.

Article 24 of the same Law supplies specific examples. It punishes direct provocation, including where the provocation is not followed by effect, to certain attacks on life, bodily integrity, sexual assaults, theft, extortion and dangerous intentional damage. It also covers provocation to discrimination, hatred or violence based on origin, ethnic or national belonging, race or religion. The relevant wording is “auront provoqué à la discrimination, à la haine ou à la violence à l’égard d’une personne ou d’un groupe de personnes à raison de leur origine ou de leur appartenance ou de leur non-appartenance à une ethnie, une nation, une race ou une religion déterminée”. A message that falsely promises entry is therefore not transformed into an Article 24 offence merely because migrants acted on it. The prosecution would need to show the prohibited call and its legally defined target.

The distinction prevents a serious category error. A person can be liable for a direct call to violence against migrants, or for knowingly helping a separately established offence, while a different person who circulates an inaccurate migration rumour may fall outside those offences. The fact that a claim has produced a public disturbance is relevant context, but it does not replace the elements of the offence. Investigators should preserve the original language, the translation used by recipients, the account’s call to action, the distribution method and the alleged underlying offence before selecting the legal qualification.

The general mental element is equally important. Article 121-3 of the Penal Code states: “Il n’y a point de crime ou de délit sans intention de le commettre”. The rule does not mean that every offence requires proof of a detailed plan. It does mean that investigators must establish the mental state required by the offence relied upon. For a direct provocation, that may involve knowledge of the message and a deliberate call to conduct. For complicity, it requires knowing assistance or instructions. For an allegation of an international influence operation, political sympathy or a shared narrative is not a substitute for proof of the accused person’s own intention.

That is why the public fact that false rumours travelled rapidly is probative but incomplete. The Associated Press fact check on the Ceuta crossing described unsubstantiated claims and misleading material surrounding the surge. The AP reconstruction of the online rumours reports that people interviewed after the crossings referred to claims that Spain’s border was open or that new arrivals would have a lawful status. Those reports can establish the existence and reach of a narrative. They cannot alone identify the first account, prove that the account holder knew the claim was false, or show that the holder intended a legally prohibited result.

The current public accusation against foreign-linked networks raises an additional attribution problem. A message can be hosted, reposted, translated, monetised, boosted by a recommendation system or copied by a user who has no knowledge of its origin. A coordinated pattern may justify an intelligence inquiry. A criminal conviction requires an evidentiary bridge between the pattern and an individual act. That bridge may include account creation records, payment instructions, administrator access, private coordination, repeated use of the same false claim after correction, or operational messages about timing and routes. It cannot be inferred simply from a nationality, a geopolitical theory or the visual similarity of several posts.

The French territorial rule is not unlimited. Article 113-2 of the Penal Code provides: “La loi pénale française est applicable aux infractions commises sur le territoire de la République” and adds that “L’infraction est réputée commise sur le territoire de la République dès lors qu’un de ses faits constitutifs a eu lieu sur ce territoire”. France cannot claim every online dispute because a video was technically visible from a French IP address. Prosecutors would need to identify a constituent act in France: for example, a French-based account issuing the call, a French-based participant providing knowing assistance, a legally relevant result in France, or another act that belongs to the offence charged.

The Cour de cassation explained the territorial principle in Criminal Chamber, 26 September 2007, appeal No. 07-83.829. Its published summary states that “il suffit, pour que l’infraction soit réputée commise sur le territoire de la République, qu’un de ses faits constitutifs ait eu lieu sur ce territoire”. The holding is useful but must not be overread. It does not say that any foreign content accessible in France automatically satisfies Article 113-2. It says that one constituent fact can be sufficient when the offence and the evidence establish what that fact is. The legal analysis must therefore begin with the chosen offence, not with the technological fact of global accessibility.

If the alleged conduct occurred entirely abroad, the French personal-jurisdiction provisions are narrower. Article 113-7 of the Penal Code extends French law to a crime or an imprisonable offence committed outside France when the victim was French at the time of the offence. The text says: “La loi pénale française est applicable à tout crime, ainsi qu’à tout délit puni d’emprisonnement, commis par un Français ou par un étranger hors du territoire de la République lorsque la victime est de nationalité française au moment de l’infraction”. The Ceuta facts publicly reported so far do not turn every person affected by a Spanish border crisis into a French victim. A French connection must be proved rather than assumed.

Where Articles 113-6 or 113-7 are used for an offence committed abroad, Article 113-8 of the Penal Code adds a procedural condition: “la poursuite des délits ne peut être exercée qu’à la requête du ministère public” and must be preceded by a victim’s complaint or an official denunciation by the country where the conduct occurred. That condition will not decide every Ceuta scenario, but it illustrates the limits of importing a foreign crisis into a French criminal docket. The nature of the alleged offence, the location of each act and the identity of the victims remain decisive.

B. When does an online messenger become an organiser or accomplice?

“Organiser” is a useful journalistic description, not a free-standing answer to criminal liability. The prosecutor must identify the principal offence and then show what the person did in relation to it. Article 121-7 of the Penal Code defines complicity in terms of a person who “sciemment, par aide ou assistance, en a facilité la préparation ou la consommation”. It also refers to a person who, through a gift, promise, threat, order, abuse of authority or power, provokes an offence or gives instructions to commit it. The word “sciemment” is central: the prosecution must prove knowing assistance, not merely participation in the same online conversation.

For a Ceuta case, the legal theory could take several different forms. One theory might allege direct provocation to violence or another offence under the 1881 Law. A second might allege that an account knowingly provided instructions or logistical assistance to a principal offence committed by identified actors. A third might concern a financial or trafficking offence under the law applicable to the conduct in Spain or Morocco, with French jurisdiction only if a French constituent act or a recognised extraterritorial basis exists. A fourth might concern racist or anti-migrant speech. These theories cannot be merged into the vague assertion that “disinformation caused migration”. Each has a different actus reus, mental element, jurisdictional test and evidentiary burden.

The causal chain also has to be individualised. A prosecutor would want to know when the account posted, which audience it reached, whether the account purchased or coordinated amplification, what the recipients understood, whether the recipients acted because of that message, and what unlawful result followed. The existence of a huge number of views does not prove that the account caused the crossing. Conversely, a small private group can be important if it contains the instructions, payment records or operational coordination that explain the public posts. The evidentiary value of a message lies in its connection to the offence, not only in its engagement count.

French corporate responsibility has its own boundary. Article 121-2 of the Penal Code states that legal entities, other than the State, are criminally responsible for offences committed on their behalf by their organs or representatives, and that this does not exclude the liability of natural persons. The key wording is “des infractions commises, pour leur compte, par leurs organes ou représentants”. This provision does not make a platform automatically criminally responsible for every user post or every algorithmic recommendation. It requires a legally defined offence and conduct attributable to the entity’s organs or representatives on its behalf.

That distinction matters when political claims point to a platform’s design. A recommendation system may have increased the reach of a false story. That fact could be relevant to a regulatory investigation, a civil claim, or proof that a particular campaign was amplified. It is not, without additional evidence, proof that the platform knowingly assisted an offence. The investigation would need to separate the platform’s ordinary technical operation from a deliberate decision by an identified person to help prepare or commit the charged offence.

The same caution applies to a person who translates or reposts a message. Translation may be neutral reporting, or it may add an instruction, remove a warning, target a new audience or coordinate a route. Reposting may be an endorsement, a warning, a fact-check or an act of assistance. The entire post, surrounding comments, private messages and subsequent conduct matter. A prosecution based only on an isolated screenshot would face obvious attribution and context challenges.

The Court of Cassation’s evidence jurisprudence reinforces that point. In Criminal Chamber, 18 June 2019, appeal No. 19-80.105, the Court held that the provisions governing technical and scientific examinations were “édictées en vue de garantir la fiabilité de la recherche et de l’administration de la preuve”. It followed that the absence of the required prosecutorial authorisation could support annulment and could be invoked by any interested party. A Ceuta investigation that uses digital forensics to attribute accounts, reconstruct deleted content or analyse devices therefore needs a procedurally valid technical foundation. A spectacular graph is not self-authenticating.

A responsible attribution exercise would test at least five possibilities separately: the first uploader, the account administrator, the person financing or directing distribution, the ordinary repeater, and the platform or service provider. It would ask which person had access to which account, who controlled the content, who received money, who gave instructions, and who knew the relevant facts at the time. It would also test innocent explanations, including copied material, compromised accounts, automated translation and users who believed an inaccurate legal claim. The more serious the accusation, the more exact the attribution must be.

The public allegation concerning Russia or Israel adds a state-attribution layer that is distinct from individual criminal liability. Intelligence services may possess classified information about influence operations. A public statement by a prime minister may justify diplomatic or counter-interference measures. A French criminal court would still need admissible evidence linking the accused person to the offence charged. The chain from “a state-linked network” to “this account holder knowingly provoked this offence” cannot be filled by political inference alone.

Finally, the underlying conduct of the migrants must not be confused with the conduct of an alleged organiser. People who moved after receiving a false message may be witnesses, victims of manipulation, or simply people who made an autonomous decision. Their presence in Ceuta does not prove that they joined a criminal conspiracy. The investigation should focus on the source and purpose of the communications, while respecting the rights of persons who crossed, requested protection or were pushed back. Criminalising the audience is not a substitute for proving the organiser’s act.

II. How could France and Spain build and test the cross-border proof?

A. Which digital evidence can investigators lawfully preserve and obtain?

The first operational step is preservation. Public posts may disappear, accounts may be renamed, stories may expire, and recommendation data may be overwritten. Investigators should record the URL, account identifier, display name, exact text, language, media file, publication time, edit history if available, replies, reposts, advertising label, audience setting and the time of capture. A screenshot is useful, but it should be accompanied by the original page or platform export, a reliable timestamp, a hash of the downloaded file and a note identifying the person or system that made the capture.

The French framework gives hosting providers a role in retaining identification information and certain removed content. Article 6 of the Law for Confidence in the Digital Economy states that access and hosting providers “détiennent et conservent les données de nature à permettre l’identification de quiconque a contribué à la création du contenu”. The same provision requires covered platforms to preserve temporarily content that they have removed or made inaccessible when it falls within the statutory categories, so that it can be made available to judicial authorities. It does not impose a general obligation to monitor every user or discover every offence in advance.

That limitation is important in a case built around a false rumour. A platform’s failure to detect an inaccurate migration claim is not automatically proof of a criminal omission. The legal question may instead be whether the platform holds identification data, whether a judicial authority made a proper request, whether the content falls within a statutory preservation duty, and whether the platform complied with an applicable order. The underlying author’s liability and the intermediary’s procedural duties must remain separate.

During a French criminal investigation, Article 60-1 of the Code of Criminal Procedure allows the prosecutor or a judicial police officer to request relevant information from a person, organisation, public administration or private body, including information from an information system or nominative data processing. The text authorises a request for information “notamment sous forme numérique”. Article 77-1-1 provides a corresponding mechanism in a preliminary investigation, including requests for digital information, subject to the statutory safeguards and the specific restrictions that apply to sensitive data.

Those provisions do not authorise an unlimited fishing expedition. A request should identify the account, time window, data sought and offence under investigation. If investigators ask for every user who searched for Ceuta, every private message mentioning Spain, or every device that viewed a video, the proportionality and legal-basis questions become acute. A narrower request for registration data and access logs for identified accounts, the original media file, deletion history, paid promotion records and administrator changes is easier to connect to the attribution issue.

Technical evidence also has to be interpreted. An IP address may identify a connection, not a person. A shared device may be used by several family members. A virtual private network, public Wi-Fi, bot service or compromised account can distort attribution. Metadata may be stripped or generated by a platform. A translation may change the apparent instruction. The digital file should therefore be compared with provider records, device evidence, payment data, witness accounts and the chronology of the real-world events. The question is not whether one dataset looks persuasive; it is whether independent sources converge on the same person and act.

Encrypted material creates another procedural issue. Article 230-1 of the Code of Criminal Procedure applies when seized or obtained data have been transformed so that the clear information cannot be accessed or understood, or when the data are protected by authentication. It permits the competent authority to appoint a qualified person to carry out technical operations “permettant d’obtenir l’accès à ces informations, leur version en clair” when necessary. The article is a tool for technical access; it does not remove the need for a lawful seizure, a defined investigation and respect for privilege, privacy and defence rights.

If relevant data are held in another European Union Member State, the European Investigation Order is the natural procedural bridge. Article 694-15 of the Code of Criminal Procedure provides that, unless the Code says otherwise, mutual assistance requests between France and another EU Member State are made through European investigation orders. Article 694-16 defines the order as a judicial decision asking the executing State to conduct investigations within a specified period to obtain or communicate evidence. It can also prevent the destruction, transformation, movement, transfer or disposal of evidence.

The order must be targeted. Under Article 694-20 of the Code of Criminal Procedure, a French prosecutor or judge may issue a European Investigation Order when it is necessary to establish, prosecute or adjudicate an offence, or enforce a sentence, and is proportionate to the rights of the suspect or convicted person. The verified text uses the formulation “nécessaire à la constatation, à la poursuite ou au jugement d’une infraction … et proportionnée”. A request to Spain for identified account data, a preserved video, a server record or a witness interview can be framed in that way. A request to map all migration-related online activity across Europe would require a much stronger justification.

The French–Spanish route is not the same as a request to Morocco, the United States or a platform whose data are outside the European Union. The European Investigation Order operates between participating Member States. For a non-EU authority, the requesting service may need an applicable treaty, a bilateral channel, a conventional mutual-assistance request or the provider’s legal-process route. The relevant route depends on where the data are held, where the provider is established, the type of data sought and the urgency. A prosecutor cannot describe an overseas platform as “in France” merely because French users accessed it.

Investigators should also seek evidence that tests causation rather than assuming it. A useful chronology would compare the first appearance of the border-opening claim, its translation into different languages, changes in the target audience, paid or coordinated amplification, corrections or official denials, travel movements and the later reports from people who crossed. It should distinguish messages that provided real-time safety information from messages that promised a legal status. It should compare the Ceuta narrative with ordinary seasonal movement, economic conditions, weather, border enforcement and reports of organised transport. The exercise is not to deny an online effect; it is to measure the effect and connect it to a legally relevant act.

Current French public guidance makes the practical point in accessible terms. Service-Public’s page on liability for content published online separates the author of illegal content, the host and the publication director. That separation should guide the Ceuta file. The author may face a direct offence; the host may have preservation or disclosure duties; the publication director may have a distinct press-law position. Treating all three as one “online organiser” would obscure the actual proof.

B. What can a suspect, platform or victim challenge?

A suspect can first challenge the legal qualification. A message may be false but not a direct provocation. It may be offensive but not directed at a protected group in the sense required by Article 24. It may be a political claim without proof of the intent required by Article 121-3. A person may have reposted it after the relevant crossing, making the alleged causal connection impossible. The defence should ask the prosecution to identify the exact offence, the principal conduct, the assistance alleged, the date and place of each act, and the factual basis for the conclusion that the accused knew what was happening.

Jurisdiction is a second line of challenge. Article 113-2 may support French jurisdiction when a constituent act occurred in France, but that fact must be demonstrated. If the prosecution relies on French nationality or a French victim for conduct abroad, Articles 113-7 and 113-8 may impose additional conditions. A French court should not become a universal forum for every platform campaign connected with Europe. The location of the author, the location of the victim, the location of the harm, the legal nature of the offence and the place of the constituent acts all need to be put on the record.

The defence can also contest digital authenticity and attribution. Who captured the post? Was the account controlled by the accused at the relevant time? Was the content edited, translated or re-uploaded? Does the provider’s record match the screenshot? Was the IP address shared, masked or assigned dynamically? Were the technical conclusions produced by a person properly appointed under the Code? The rule recalled in appeal No. 19-80.105 is especially relevant: the procedural safeguards for technical examinations protect “la fiabilité de la recherche et de l’administration de la preuve”. Reliability is not a decorative concern; it can determine whether the evidence enters the case.

Proportionality is a third issue. The public interest in investigating an unprecedented border movement is strong. It does not authorise indiscriminate access to private communications, location histories or the accounts of every person who discussed the crisis. A request should be limited by account, period, category of data and offence. Where privileged communications, journalists’ sources, minors’ data or asylum information may be involved, the authority must apply the specific safeguards. The order should explain why the less intrusive source is insufficient.

Platforms have their own arguments. The LCEN requires identification data and, in defined circumstances, preservation of removed content. It does not create a general obligation to monitor all information or guarantee that every false claim will be removed before it spreads. The Constitutional Council’s 2020-801 DC decision rejected an overbroad and disproportionate removal mechanism, stressing the difficulty of classifying online content under a short deadline. That decision does not immunise platforms from lawful orders. It does prevent a conclusion that failure to remove a disputed message is, by itself, proof of criminal assistance.

A platform or account holder may also need to distinguish lawful disclosure from a voluntary overproduction of data. The judicial request should identify the legal basis, the requested fields, the period, the relevant account and the safeguards. If a European Investigation Order is used, the executing authority may examine recognition, execution, necessity, proportionality and the rights of the person affected under the applicable framework. If data are requested from a third country, the provider’s disclosure process may impose additional requirements. A cross-border file is not procedurally stronger merely because several authorities repeat the same demand.

Victims and residents can preserve evidence and seek redress without making unsupported accusations. A person targeted by racist messages should save the original URL, date, account identifier, translations, threats, reposts and any resulting harm. A business or public official whose reputation or safety was affected may need a different civil or criminal route. Article 1240 of the Civil Code states: “Tout fait quelconque de l’homme, qui cause à autrui un dommage, oblige celui par la faute duquel il est arrivé à le réparer”. The provision can support a civil-liability analysis when fault, damage and causation are established; it does not eliminate the need to prove those elements.

For victims who are not French and conduct that occurred in Spain, the French civil or criminal route will depend on the applicable jurisdictional rules and the connecting factors. A complaint in France should state why the French authority is competent and identify the specific offence or damage. A complaint that simply repeats the political attribution to a foreign state may be redirected, dismissed or treated as intelligence rather than evidence. The strongest file starts with preserved content and a precise account of the harm.

The status of the migrants also requires legal care. The public record includes people who returned to Morocco, people who remained in Ceuta, persons seeking asylum and persons who may have been victims of violence or exploitation. Their statements may help establish what the online messages promised and how they were understood. They should not be treated as proof of an organiser’s intention without corroboration. Their immigration status, vulnerability, age and protection claims can affect how evidence is collected and how a witness is interviewed.

The geopolitical allegation should be tested in the same way. If Spanish or European services identify coordinated foreign information manipulation, the relevant evidence may include infrastructure links, funding, administrator accounts, operational instructions, repeated deployment across channels and a demonstrable purpose. The existence of a narrative that benefits a foreign actor is not enough. Nor is a content label such as “pro-Russian” or “pro-Israeli” a substitute for attribution. The legal file must distinguish a state’s responsibility, an organisation’s responsibility, an account holder’s responsibility and an ordinary user’s conduct.

For counsel reviewing a new file, the immediate checklist is practical. Preserve before requesting deletion. Obtain the original language and a controlled translation. Record the chain of custody. Identify the offence before requesting data. Map every act to a country and date. Test whether the message was a call, a report, a warning or an instruction. Separate the author from the platform. Request only data that can answer the attribution and causation questions. Use a European Investigation Order for a defined evidentiary measure in another Member State. Seek human review before publishing or repeating an allegation of foreign direction.

The existing Ceuta analysis on asylum in France and possible transfer procedures addresses a different question: what a person who passed through Spain may face when seeking protection in France. It should be used as an internal procedural complement, not as proof that the border rumour was criminally organised. Keeping those issues separate is essential to avoid turning a migration-law question into an unsupported criminal accusation.

Conclusion

The Ceuta crisis is a credible subject for a cross-border criminal and evidence inquiry because online rumours, a mass movement, alleged foreign amplification and multiple jurisdictions converged in a matter of days. The public facts nevertheless stop short of proving a French offence. France would need a defined criminal qualification, a constituent act connecting the case to French territory or another recognised basis of jurisdiction, evidence of the required intent, and a reliable chain linking a particular person to the relevant communication and result.

The legal route is therefore not “disinformation equals crime”. It is a sequence: identify the message, identify the offence, preserve the digital record, attribute the account, test causation, choose the competent jurisdiction and obtain foreign evidence through a proportionate procedure. Articles 23 and 24 of the 1881 Law may apply to direct provocation or hate-based calls; Articles 121-3 and 121-7 of the Penal Code require the mental and assistance elements; Articles 60-1, 77-1-1, 230-1 and 694-15 to 694-20 provide procedural tools with safeguards. Political statements can trigger scrutiny, but only verified evidence can support criminal responsibility.

For an international audience, that French procedural distinction is the practical answer. A coordinated online campaign may be real, yet attribution and jurisdiction remain questions for proof. A careful investigation can protect the public, preserve evidence and hold an actual organiser to account without treating migrants, platforms or foreign states as guilty by association.

📄 Circulaire officielle