Anthropic's new classroom assistant, Claude for Teachers, promises to save educators hours of planning time by generating lessons and analyzing rosters. However, data privacy experts warn that uploading student files to this free tool violates federal law and exposes teachers to legal liability. By prompting educators to upload diagnostic reports and attendance sheets, the platform creates a compliance risk that violates school district policies.
What Happened
Anthropic recently launched Claude for Teachers, a free version of its artificial intelligence assistant for U.S. educators. The tool links to the Learning Commons portal, which contains open-source curricula like OpenSciEd and Illustrative Mathematics. While we previously reported that classroom AI use is growing rapidly, this release worries privacy advocates. Anthropic's marketing encourages teachers to upload "folders of data," including rosters, grade diagnostics, and notes. Although Anthropic's K-12 Data Protection Addendum claims to align with the Family Educational Rights and Privacy Act (FERPA), legal analysts say individual teachers cannot sign these terms for their schools.
The Bigger Picture
To handle student data legally under FERPA, an education technology vendor must meet the "school official" exception. As explained by Promise Legal, a vendor only qualifies if it is under the "direct control" of the school district, which requires a formal, board-approved contract. Individual teachers cannot designate a company as a school official on their own. When a teacher signs up for a free tool, the school district technically violates FERPA because there is no district oversight.
The risk is higher for special education records. Uploading information from an Individualized Education Program (IEP) or a Section 504 plan violates protections under the Individuals with Disabilities Education Act (IDEA). According to the Special Education Law Blog, entering raw diagnostic data into AI models compromises confidentiality, even if the company says it does not train its models on that data. School districts, rather than the AI vendors, keep all legal liability if these tools generate incorrect educational goals.
What This Means for Families
For parents, this means sensitive grades, diagnostic evaluations, and medical accommodations may go into third-party databases without consent. For teachers, using these tools without approval carries severe consequences. Many free AI services include indemnification clauses. Under these terms, the teacher, rather than the school district, faces personal and financial liability for data breaches. This puts educators between classroom administrative workloads and federal privacy laws. Some states are building centralized networks to secure student data. For example, the Region 4 Education Service Center launched the Texas Education Exchange to run secure, district-controlled data pipelines, removing the need for teachers to use unvetted tools.
What You Can Do
- Check your district's pre-approved list of software vendors with active Data Protection Agreements (DPAs) before using any AI assistant.
- Remove all personal details, rosters, grades, and diagnostics when using AI tools, and use placeholders like "Student A" instead.
- Push for formal district evaluations of tools like Claude for Teachers so schools can set up safe, official agreements.