Norway's National Tiered Restriction on Generative AI in Primary and Secondary Schools
Norway
Norway's government announced on 19 June 2026 that pupils aged 6-13 will be barred from using generative AI in school …
United States of America · Mountain View · See the United States of America profile · See the Mountain View profile
Top 98% 13/100 · Ask Evidence Copilot about this practice
Google switched on Gemini by default for all K-12 students in Google Classroom in 2026 with only days of notice to schools. An independent safety audit found it misses warning signs across 13 mental-health conditions and treats a sixth-grader like a high-school senior.
In 2026 Google enabled its Gemini chatbot by default inside Google Classroom, which serves more than 150 million students and teachers worldwide, extending access from users 18 and older to every K-12 student unless a school proactively moved under-18 accounts into a separate organisational unit to opt out. Multiple US school leaders, including a principal quoted by Education Week, said they learned of the change from news coverage rather than from Google or their own district, and education researcher Justin Reich noted that schools were again forced to react 'whenever the tech industry decides we have to deal with it.'
An independent risk assessment by Common Sense Media's Youth AI Safety Institute tested Gemini K-12 against 13 mental-health topics (including anxiety, depression, eating disorders, mania, psychosis and self-harm) and rated it high risk. The audit found the tool 'fails to recognize warning signs' in most categories, 'easily backs down when challenged,' and in some tests issued a safety warning about risky behaviour before following it with step-by-step instructions that undermined that same warning. Because the system treats all under-18 users identically, a sixth-grader receives the same response as a high-school senior, at an output reading level the assessment put around 11th grade -- too advanced for most K-12 users. Testers also found the tool would complete full homework assignments and essays even in a 'Guided Learning' mode meant to scaffold rather than do the work.
By contrast, the assessment noted that Microsoft Copilot, OpenAI's ChatGPT and Anthropic's Claude all require stricter age verification or manual administrator activation before reaching minors. The episode is documented here as a cautionary case: default-on deployment at massive scale, without proportionate safeguards or adequate notice, is a documented governance and child-safety failure mode for AI in education, not a practice to replicate.
Read the full analysis: https://institute.commonsensemedia.org/risk-assessments/gemini-k-12
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
Norway's government announced on 19 June 2026 that pupils aged 6-13 will be barred from using generative AI in school …
United States of America
Saint Paul Public Schools' 2025 contract with the St. Paul Federation of Educators bars adverse employment action based solely on …
El Salvador
El Salvador announced a two-year plan to deploy xAI's Grok as a digital tutor to 5,000+ public schools reaching a …
United States of America
After MagicSchool AI's companion chatbot "Raina" appeared on student iPads, 1,100+ Bend-La Pine parents petitioned the Oregon district; the vendor …
Open full copilot Grounded in cited practices — always check the sources.