Anthropic agreement sets student data rules for school AI
The American Federation of Teachers (AFT) and United Federation of Teachers (UFT) announced a legally binding agreement with Anthropic on November 10, 2026. It sets enforceable safety, privacy and transparency standards for Anthropic’s AI in thousands of U.S. K-12 schools.

Key takeaways · 4
- 01
Existing district agreements for Claude for Teachers are covered automatically; future licensing contracts require districts to opt in.
- 02
Schools retain control over data retention and deletion, and Anthropic must limit information use to authorized educational purposes.
- 03
Student, educator and customer data cannot be used to train or improve models, apart from specified research and safety-related data.
- 04
District policies and staff training must ensure human review of AI-informed grading, discipline and class-placement decisions.
What the agreement covers
The AFT and UFT said the legally binding agreement makes safety, privacy and transparency standards enforceable in thousands of U.S. K-12 schools.[1] It took effect on the day of the November 10, 2026 announcement, after Anthropic had already committed to it.[1] The agreement automatically applies to Anthropic’s existing school-district agreements for Claude for Teachers.[1] It covers Claude for Teachers, but does not necessarily cover education products built by other companies using Claude models.[3]
Data use and safeguards
The agreement limits student and educator information to authorized educational purposes, bars its sale or use for targeted advertising, and prohibits using student, educator or customer data to train or improve AI models, with specified research and safety-related exceptions.[1][3] Schools retain control over how long their data is kept and when it is deleted.[2] Anthropic must alert customers to data breaches and mitigate them, including after a contract ends.[3]
Educators remain responsible for decisions
The agreement leaves instructional and other consequential decisions about students to educators, rather than AI systems.[1] Districts must adopt policies and train staff to ensure people review AI-informed grading, discipline and class-placement decisions.[3] Schools, educators and families are also to receive plain-language information about how the products work, the data they collect, their shortcomings and their safeguards.[1]
Existing deals and future adoption
Districts do not need to opt in for the agreement to apply to their existing Anthropic agreements for Claude for Teachers, but they must opt in for future licensing contracts.[1][3] The AFT and UFT had previously announced Microsoft’s adoption of the AI safety and privacy agreement in September 2026, and Microsoft’s agreement took effect for existing contracts in November 2026.[2][3] The unions said they are working to hold AI companies accountable because the federal government has not stepped in, and the AFT has described federal and state AI regulation as almost entirely absent.[1] The union hopes the standard will eventually be embedded in law.[2]
School and district teams evaluating AI tools can use the agreement’s provisions as concrete checks for data handling, safeguards, transparency and human review. The distinction between existing and future contracts makes it important to verify whether a district’s agreement is covered automatically or requires an opt-in.
Why it matters
AI training for your team
kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.
How this developed
13 November 2026
Anthropic agreement sets student data rules for school AI
Sources
- AFT, UFT and Anthropic Sign Binding AI Safety and Privacy Agreement for Schools | American Federation of Teachersaft.org
- Anthropic enters student AI privacy agreement with teachers unions | K-12 Divek12dive.com
- Anthropic Agrees to Protections for Teachers Who Use Claudeedweek.org
- Introducing Claude for Teachers \\ Anthropicanthropic.com