Binding law — in force AI-adjacent law
Label influencer content that shows an AI-produced face or silhouette 'Images virtuelles' (France, Law 2023-451 Art. 5)
In France, content that a person exercising commercial influence (anyone who, for payment, uses their standing with an audience to promote goods, services or a cause online, Art. 1) communicates must carry the mention 'Images virtuelles' when it includes images produced by an AI process to depict a face or a silhouette (Law 2023-451 Art. 5 I 2°, in the version in force since 2024-11-08). The mention must be clear, legible and understandable on every medium; an equivalent mention suited to the activity and the format may replace it. Breach is an offence punishable by one year's imprisonment and a EUR 4,500 fine (Art. 5 III). Persons established in another EEA State are outside Art. 5 save after the e-commerce directive procedure (Art. 5-1 II). Detect influencer-content generators that produce faces or bodies with no virtual-image mention.
Trust and provenance not reviewed by a lawyer · audit-grade · source verified 5 Oct 2026 · release 2026.10.05
- Lane
- Binding law — in force In force: applies since 8 Nov 2024
- Official source
- Loi n° 2023-451 du 9 juin 2023, art. 5, I, 2° (« Images virtuelles ») · captured 5 Oct 2026 · anchor hash (SHA-256)
ca4ef1ca3b79…· 8 more anchors in the data release - Verification
- Quoted text found word for word in the captured official document (5 Oct 2026). Source last verified 5 Oct 2026: checked against the captured official document; not in the weekly watcher's list; checked against the captured document.
- Data release
- Data release 2026.10.05, data as of 5 Oct 2026, schema 0.3.11. This page also reflects corpus changes made after that release; they ship in the next one.
- Legal review
- Not reviewed by a lawyer. TwinEthos derived this rule from the official text it cites: treat it as research to check against that text; it is not legal advice. No TwinEthos rule has been legally reviewed yet. Open questions for counsel on this rule: 1.
- Audit standard
- Audit-grade: meets all 10 checks of the TwinEthos audit standard that apply to it. The audit standard is TwinEthos's own quality bar for provenance, dates, applicability, detectors, fixtures, remediation and licences; it is not a legal review.
- Detectors
1 detector (code pattern), experimental: written from the rule's text and not yet measured for precision on real code, so treat a hit as a lead to verify.
Known limits:
- Faces produced in an external design tool and uploaded as finished files
- Equivalent mentions in other languages than French and English ('Images virtuelles', virtual images, AI-generated image)
- The mention may be added by a rendering service, a caption template or the publishing platform in another file; follow the output to the publish call before reporting. Product shots, backgrounds and other images with no…
Who it applies to
- Duty falls on: any person
- Natural or legal persons who, for consideration, use their reputation with their audience to communicate content online promoting goods, services or a cause (commercial influence, Art. 1), for content that includes images produced by any AI process to depict a face or a silhouette. Applies from 2024-11-08 (Art. 5 in the version in force since that date). Not applicable to persons established in another EEA State save after the Directive 2000/31/EC procedure (Art. 5-1 II). The guard sits in the image or video generator, or the publishing tool, of an influencer-content product. Whether a post is paid promotion, and where the influencer is established, are facts outside the code.
- Not covered:
- Persons established in another State party to the European Economic Area: Art. 5 does not apply to them, unless, after the procedure of Art. 3(4)-(5) of Directive 2000/31/EC, the administrative authority tells the person which provisions apply (Art. 5-1 II)
- Whether it applies depends on facts outside the code; a person has to decide.
The guard to add
Mark every AI-produced face or silhouette in influencer content and publish it with a legible 'Images virtuelles' (or equivalent) mention.
Where the product calls an image or video model that produces a person, a face, a body or a virtual influencer (text-to-image, face swap, virtual try-on, avatar video), the output is flagged ai_face_or_silhouette; the renderer then overlays a legible mention on the image or video ('Images virtuelles' for France, a configured equivalent elsewhere) and the publisher adds it to the caption, and the publish or export call refuses a flagged output that lacks it. Retouching that only slims or reshapes a real photo is a separate label ('Images retouchées' in France) and is not covered by this control.
Where it goes: 9 AI output handling, 1 application source code, 14 user-facing text.
What this provision adds:
- The mention reads 'Images virtuelles' or an equivalent suited to the activity and the format, and is clear, legible and understandable on every medium used.
Example (Python creator studio + OpenAI images + Pillow), before:
img = client.images.generate(model='gpt-image-1', prompt=f'{persona} wearing {product}, studio portrait')
post = publish_to_instagram(decode(img), caption=caption)After:
MENTION = 'Images virtuelles'
img = decode(client.images.generate(model='gpt-image-1', prompt=f'{persona} wearing {product}, studio portrait'))
img = draw_label(img, MENTION, min_height_ratio=0.04) # legible on the image itself
post = publish_to_instagram(img, caption=f'{caption}\n{MENTION}', ai_face_or_silhouette=True)Control: Commercial-influence content showing an AI-produced face or silhouette is published without a virtual-image mention. The same guard addresses 1 item with binding law in 1 jurisdiction. Engineering guidance, not legal advice.
Rule id fr-law-2023-451.ai-face-or-silhouette-images-virtuelles · review status: primary source derived