Skip to content

Regulation Tracker · Country · Europe

France

Extensivecivil lawupdated 2026-09-04verified 2026-09-04

EU-level rules also apply — see the European Union record →

As of 4 September 2026 France regulates legal AI through professional deontology and judicial charters rather than a court rule. The CNB adopted a deontological AI guide on 17 March 2026; the Conseil d'État's charter of 11 December 2025 binds every member of the administrative jurisdiction; article L. 111-13 COJ criminally bans profiling judges; administrative courts have begun censuring hallucinated citations.

As of 2026-09-04, France has 4 binding rules, official guidance in 12 categories, 0 proposals and 2 positions set only by case law across the 20 categories in the map. Binding rules cover AI barred from deciding cases, consumer AI legal services, horizontal AI statute, mandatory AI training. In 2 categories we found nothing; in 0 the position is not yet determined.

How to read this page
  • Binding rule Binding rule means a rule in force from a court, statute, regulator or equivalent authority.
  • Official guidance Official guidance means an official non-binding position from a judiciary, regulator, bar or public body.
  • Proposed / consultation Proposed / consultation means a public bill, draft rule or consultation that is not yet in force.
  • Set by case law Set by case law means no rule was found, but courts have set the position in decisions.
  • Nothing found Nothing found means the category was checked and no rule, guidance, proposal or case-law position was found.
  • Not yet determined Not yet determined means research did not establish a reliable position yet.

The rules, category by category

AI barred from deciding cases

Binding rule

verified 2026-09-04

Article 47 of the loi Informatique et Libertés (6 Jan 1978) forbids any judicial decision involving an assessment of a person's behaviour being founded on automated processing profiling that person. The Conseil d'État charter restates it: 'l'IA est un outil qui jamais ne décide' — AI may not interpret a rule, establish or assess facts, apply law to facts or propose an outcome. Separately, article L. 111-13 COJ (inserted by loi 2019-222, art. 33) makes it a criminal offence to reuse judges' and clerks' identity data to evaluate, analyse, compare or predict their professional practices — France's unique ban on judge analytics.

Consumer AI legal services

Binding rule

verified 2026-09-04

Articles 54 to 56 of loi no. 71-1130 of 31 Dec 1971 reserve paid legal consultation and drafting of acts to regulated professionals; they are technology-neutral and catch consumer AI legal products. No AI-specific French ruling found as of 4 September 2026.

Horizontal AI statute

Binding rule

verified 2026-09-04

The EU AI Act applies directly; Annex III point 8(a) classifies AI used by or for judicial authorities to research/interpret facts and law as high-risk. France designated the CNIL as lead AI Act authority and the DGCCRF as market-surveillance coordinator and single point of contact, with about fifteen sectoral regulators (Arcom, ACPR, AMF, ANSM, HAS).

Mandatory AI training

Binding rule

verified 2026-09-04

Article 4 of the EU AI Act has required deployers — including law firms and courts — to ensure a sufficient level of AI literacy among staff since 2 Feb 2025. No French mandatory AI CLE: the CNB's Skilia/Lefebvre Dalloz course (10,000+ enrolments, open to 2027) is voluntary, and the Ministry of Justice offers non-compulsory modules to court staff.

Duty to verify AI output

Official guidance

verified 2026-09-04

Two official texts impose it. The Conseil d'État charter (11 Dec 2025) requires personnel to 'assurer systématiquement une vérification humaine' and never to assume AI output is true or complete. The CNB deontological guide (17 Mar 2026) makes the lawyer keep 'la maîtrise intellectuelle' of the work. Administrative courts have applied the duty (TA Orléans, 29 Dec 2025).

Judges' use of AI

Official guidance

verified 2026-09-04

Conseil d'État charter of 11 Dec 2025 applies to every member, magistrate and staff member of the Conseil d'État, administrative courts of appeal, tribunaux administratifs, the CNDA and the Tribunal du stationnement payant. Seven principles; access to public chatbots is deliberately not blocked but is tightly conditioned. No equivalent charter for the ordre judiciaire yet — the Ministry of Justice says one is in preparation.

Client data in AI tools

Official guidance

verified 2026-09-04

CNB guide (17 Mar 2026) puts protection of the secret professionnel and GDPR compliance first. The Conseil d'État charter tells court personnel to give up using external AI where data security and confidentiality cannot be assured: giving information to a public chatbot 'équivaut à la publier sur internet'.

Technological competence

Official guidance

verified 2026-09-04

The Règlement intérieur national (art. 1.3) already requires compétence, dévouement, diligence et prudence; the CNB guide of 17 Mar 2026 applies competence, prudence and independence expressly to generative AI. No amendment to the RIN naming AI has been made.

Bar guidance on generative AI

Official guidance

verified 2026-09-04

Two CNB documents: the 'Guide pratique — Utilisation des systèmes d'intelligence artificielle générative' (1st ed., September 2024, 42 pp.) and the 'Guide pratique — La déontologie et l'intelligence artificielle' adopted at the assemblée générale of 17 March 2026. The CNB also republished the CCBE guide and runs a free Skilia/Lefebvre Dalloz AI course open until 2027.

Courts' own AI deployment

Official guidance

verified 2026-09-04

Administrative jurisdiction: the Conseil d'État charter covers internal SIA and excludes high-risk systems, limiting internal projects to low-risk tasks (anonymisation, research support). Judicial order: the Ministry of Justice report 'L'IA au service de la justice' (Aug 2025) and a sovereign internal generative-AI assistant for magistrates and staff, with 'Mon Assistant Pénal' and 'Mon Assistant Civil' built with the Paris Court of Appeal.

Arbitration and ADR

Official guidance

verified 2026-09-04

No French statute or institutional rule. Paris-seated ICC arbitration is covered by the ICC Note to Parties and Arbitral Tribunals (1 Jun 2026, paras 76–80) and the ICC Commission task force on AI; the Ciarb Guideline on the Use of AI in Arbitration (Mar 2025, updated Sept 2025) is the usual reference text. See the international map.

Vendor due diligence

Official guidance

verified 2026-09-04

The CNB published a 'grille de sélection des outils d'IA' alongside the 2024 practical guide, scoring data sovereignty, security, functionality, ethical compliance and cost. The Ministry of Justice states a preference for solutions hosted on national territory.

Sanctions for AI misuse

Set by case law

verified 2026-09-04

At least eight French decisions since December 2025 have flagged generative-AI misuse. TA Orléans (29 Dec 2025, no. 2506461) censured a represented applicant's counsel for about fifteen wholly fabricated references; TJ Périgueux (pôle social, 18 Dec 2025) found fictitious case references in submissions. So far the courts have issued judicial warnings and rejected the submissions; no bar disciplinary sanction and no article 32-1 CPC amende civile has been reported.

Disclosure of AI use in filings

Nothing found

verified 2026-09-04

No court rule, decree or practice direction requires parties or counsel to disclose or certify AI use in filings as of 4 September 2026. Administrative courts have criticised undisclosed AI use in submissions but imposed no disclosure obligation.

AI-generated evidence

Nothing found

verified 2026-09-04

No French evidence rule or judicial guidance on AI-generated or deepfake evidence found as of 4 September 2026.

Official documents recorded for France (4)

Court incidents on the record in France

all incidents →

Cite this page

SafeLegalAI Legal AI Regulation Tracker, "France" (updated 2026-09-04, verified 2026-09-04), https://safelegalai.com/regulation/country/fr (accessed 2026-09-04). Data: CC BY 4.0.

SafeLegalAI is a research publication by Cognesio LLP, not a law firm. Nothing here is legal advice, and no lawyer–client relationship arises from reading it. Rules change; always check the official document linked on each record and take advice on your own situation. Researched and drafted with AI assistance; verified against primary sources and edited by Cognesio LLP. The linked official documents are the record — our summaries are not the law and are not legal advice.