Norway's National AI Regulatory Sandbox — Testing High-Risk and Generative AI Before Deployment
Norway
Since 2021, Norway's data protection authority has run about 20 time-boxed AI sandbox projects with public bodies — from NAV's …
Uruguay · Montevideo · See the Uruguay profile · See the Montevideo profile
Top 83% 40/100 · Ask Evidence Copilot about this practice
Decree 276/025 (Dec 2025) lets AGESIC approve temporary regulatory sandboxes and data spaces for AI projects, capped at two years, with impact assessments and data-protection sign-off; no projects had been reported at launch.
Uruguay's Decree 276/025, promulgated on 2 December 2025 and published on 10 December 2025, implements Articles 74 and 75 of Law 20.212 by creating a minimum governance scheme for 'controlled testing environments' (regulatory sandboxes) and 'data spaces' for innovative projects, especially in data use and artificial intelligence.
AGESIC approves each environment by reasoned resolution. A sandbox may last at most two years, and any regulatory exception granted cannot outlast it. Applicants must submit a timeline, the persons affected, data-protection impact assessments, an estimate of possible harms with remedies, and a public-interest justification where relevant. Use of personal data requires a favourable opinion from the personal-data regulator (URCDP).
A Technical Evaluation Committee chaired by AGESIC, with the URCDP, the Ministry of Industry and the national research and innovation agency, manages applications, monitoring and future regulatory recommendations. Principles of equity, non-discrimination, accountability, transparency, audit and safe innovation are written into the decree.
Evidence caveat: the sources describe the legal framework only. Press coverage at launch mentioned no admitted projects, outcomes or published evaluation, so this entry records a governance design to watch, not a demonstrated result.
Read the full analysis: https://www.impo.com.uy/bases/decretos/276-2025
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
Norway
Since 2021, Norway's data protection authority has run about 20 time-boxed AI sandbox projects with public bodies — from NAV's …
Australia
Canberra required 94 federal agencies to publish AI-use transparency statements by 28 Feb 2025. The Digital Transformation Agency says all …
Ireland
Ireland's May 2025 guidelines set seven principles, a Decision Framework and a 'Responsible AI Canvas' for public bodies considering AI, …
Latvia
Latvia's newly created AI Centre ran the country's first AI regulatory sandbox call: 10 applications, 3 projects selected, free access …
Open full copilot Grounded in cited practices — always check the sources.