Page traduite et vérifiée automatiquement. La version anglaise fait foi. Certaines parties restent en anglais le temps que leur traduction soit vérifiée. Voir en anglais · Signaler une erreur de traduction
Loi du Colorado sur l’IA
Colorado Artificial Intelligence Act (SB 24-205)
La première loi générale d'un État américain sur la discrimination algorithmique, inspirée en partie d'approches fondées sur les risques comme le règlement européen sur l'IA.
The law in brief
The Colorado AI Act (SB 24-205) was the first comprehensive US state law aimed at algorithmic discrimination. It puts duties on developers and deployers of high-risk AI systems, those that make, or are a substantial factor in making, consequential decisions about people.
Its status is unsettled. Its effective date was moved to 30 June 2026, and in April 2026 a federal court stayed its enforcement in a pending lawsuit, while the legislature has been revisiting the framework. The obligations below are the law as enacted; check Regulatory Watch before relying on dates.
À qui elle s’applique
- Developers of high-risk AI systems doing business in Colorado.
- Deployers that use high-risk AI systems to make consequential decisions about Colorado consumers.
- Consequential decisions are those with a material effect on access to, or the cost or terms of, education, employment, financial or lending services, essential government services, health care, housing, insurance or legal services.
- Some small deployers (fewer than 50 employees, meeting conditions) are exempt from parts of the deployer duties.
What it requires
A duty of reasonable care
Developers and deployers must use reasonable care to protect consumers from known or reasonably foreseeable risks of algorithmic discrimination.
Deployer risk management and impact assessments
Deployers maintain a risk management policy and program, and complete an impact assessment for each high-risk system at least annually and after significant changes.
Tell consumers, explain, and allow appeal
Notify consumers before a high-risk system is used in a consequential decision about them, and after an adverse decision explain the reasons and offer correction and appeal.
Developer documentation and disclosure
Developers give deployers the documentation needed for impact assessments, publish a statement on their high-risk systems, and tell the Attorney General and known deployers of discovered risks of algorithmic discrimination within 90 days.
Droits des personnes
Consumers must be told when a high-risk AI system is used to make a consequential decision about them. After an adverse decision they are entitled to an explanation of the principal reasons, a chance to correct incorrect personal data, and an appeal with human review where technically feasible.
Enforcement and penalties
Application assurée uniquement par le procureur général du Colorado, les violations étant traitées comme des pratiques commerciales trompeuses au sens du Colorado Consumer Protection Act ; il n’existe aucun droit d’action privé. Le respect d’un cadre reconnu tel que le cadre de gestion des risques liés à l’IA du NIST ou la norme ISO/IEC 42001 permet d’invoquer un moyen de défense.
What's changing
Stayed and under revision. Enforcement was stayed by a federal court in April 2026, and further legislative changes have been under consideration. Treat the duties below as the direction of travel, and check Regulatory Watch for the current position.
What to do first
- Inventory AI used in consequential decisions about Colorado consumers.
- Adopt a risk management program aligned to the NIST AI RMF or ISO/IEC 42001; it helps under any version of the law.
- Prepare impact assessments for high-risk systems.
- Draft consumer notices and an adverse-decision explanation and appeal process.
- Watch the court case and the legislature before committing to dates.
Checked against the official text on 28 September 2026. Quotations are the operative words of the law, linked to the article they come from; the official text is the authority. This brief is written by the Association for practitioners and is not legal advice.
Évolutions récentes
Rien de signalé pour l’instant. Regulatory Watch recense les nouvelles règles, les mesures d’application et les décisions de justice au fur et à mesure.
Sources
- Colorado SB 24-205, protection des consommateurs en matière d’intelligence artificielle Assemblée générale du Colorado · Texte officiel ou régulateur
- Champ d’application de la loi du Colorado sur l’IA Privacy World · Cabinet d’avocats ou analyste
Bibliothèque des réglementations
Toutes les lois des États américains sur la vie privée, sur une seule carte
24 États disposent d’une loi générale sur la vie privée dans la bibliothèque, aux côtés des lois de l’UE, du Royaume-Uni, des lois fédérales américaines et des lois du reste du monde : 102 au total, chacune expliquée en termes simples avec ses sources. Sélectionnez un État pour voir ce qui s’y applique.
- En vigueur
- Entrée en vigueur prochaine
- Lois sectorielles