Algorithms on Nice's cameras: the carnival face test and the school-gate system the courts stopped
This overview was generated with AI from the public sources listed below and checked against them before publication. It has not had a complete human fact-check — treat it as a starting point and follow the sources.
How this atlas is made · What “last verified” means · What “source-checked” means
During the February 2019 carnival Nice ran a facial-recognition test on its public-road cameras at one entrance to the carnival zone, matching passers-by against fictional watchlists of missing, wanted and banned persons; 50 volunteers, mostly municipal staff, were enrolled and about 5,000 people passed the gate. The city reported a success in June 2019; in July 2019 the CNIL asked for objective measurement of the algorithm's error rates, image-quality conditions and bias risks, which the city said it lacked the perspective to supply. The system that actually ran was duller and lasted longer: from 2020 an algorithm analysed the cameras filming 144 school entrances and raised an alert whenever a vehicle stayed parked there more than five minutes. After an on-site inspection on 5 April 2023 and a formal notice on 13 November 2024, the CNIL held on 15 May 2025 that the treatment was not authorised; Nice challenged the opinion and lost.
what does this mean?
The strongest source is an independent non-academic evaluation: an audit office, a government or EU review, or an NGO. It grades who did the looking — not how well the project went. All four grades →📊 Impact
On 30 January 2026 the Conseil d'État (decision n° 506370) rejected Nice's appeal and confirmed the CNIL's reading: article L. 251-2 of the code de la sécurité intérieure permits video surveillance of public roads but "cannot, in its silence, be interpreted as authorising algorithmic treatments allowing systematic and automated analysis of the images collected", and no other text authorises it. The ruling applies to every French municipality, not only Nice.
🎓 Lesson
Consent could not carry the 2019 test either: the CNIL accepted it only as a real-conditions trial on volunteers "without operational consequence", and in the same November 2019 position paper it declared facial recognition for access control at schools off-limits — a line the Marseille administrative court drew in law on 27 February 2020 by annulling the PACA region's biometric gates at the Lycée Les Eucalyptus in Nice and the Lycée Ampère in Marseille. Those gates were the region's project, not the city's; what belonged to Nice was the camera estate underneath.
Sources
Show 2 more sourcesShow fewer
Continue exploring
Compare this project
Facial recognition on public cameras: what stopped it?
In four cities the cameras were already there and the face-matching layer was the question. Each was halted — by a different institution, on different grounds, and one of them was never switched on at all.
| Project | Since | Status | Outcome | Evidence | Run by | Cost | What stopped it |
|---|---|---|---|---|---|---|---|
| Surveillance Technology Ordinance & Facial Recognition BanSan Francisco, 🇺🇸 | 2019 | Live | Scaled back | Official | City government | not public | The city stopped itself: an ordinance barred its own departments from face recognition and made every other surveillance tool pass a council vote. |
| Buenos Aires's Fugitive Facial Recognition System, stopped by the courtsBuenos Aires, 🇦🇷 | 2019 | Suspended | Failed | Independent | City government | not public | A court stopped it — on the city's own security law and the oversight bodies it never convened, rather than primarily on data protection. |
| Algorithms on Nice's cameras: the carnival face test and the school-gate system the courts stoppedNice, 🇫🇷 | 2019 | Suspended | Failed | Independent | City government | not public | The regulator asked the city to measure its own error rates, the city could not, and the highest administrative court then ended the practice for every French municipality. |
| 112 Downtown Cameras, and the Ten Face-Matching Ones That Never Switched OnCampo Grande, 🇧🇷 | 2024 | Live | Scaled back | Media | City government | not public | Nothing stopped it, because it never started: the municipal guard holds no image database, and the data-sharing agreement was never signed. |
Explore the wider context
📎 Cite this project
1001 Smart Cities (2026). “Algorithms on Nice's cameras: the carnival face test and the school-gate system the courts stopped” — Nice, France. The Smart City Atlas. https://1001smartcities.org/projects/nice-camera-algorithms/ (last verified 2026-08-10). Data: CC BY 4.0.
The underlying data is free to reuse with attribution — see the open data page.