Anthropic’s Claude for Teachers lets K–12 educators generate state-aligned lessons by tapping Learning Commons and established curricula. While the product promises strong privacy protections, marketing that urges teachers to upload class data conflicts with FERPA and CfT’s own terms. The service’s contract asks teachers to designate Anthropic as a "school official" and to indemnify the company, shifting risk to educators. The authors call for district-controlled data governance and collaboration between AI labs and school systems.
Opinion: Claude for Teachers Offers Promise — But Its Terms Could Expose Educators to Student-Data Risks
Anthropic recently launched Claude for Teachers, a free version of its Claude AI aimed at K–12 U.S. educators who register with a school email and confirm classroom use. The tool integrates with Learning Commons and draws on proven curricula such as OpenSciEd and Illustrative Mathematics, enabling teachers to generate state-aligned lessons and instructional materials quickly.
Why Teachers Find It Attractive
Given the heavy planning burden on teachers, Claude for Teachers (CfT) is immediately appealing: it can produce lesson plans, align content to state standards and streamline differentiation. Anthropic also advertises strong privacy features — promising not to use teacher inputs to train models and offering encryption, deletion rights, audits and breach notifications — which may reassure educators wary of consumer-grade AI tools.
Where Marketing Collides With School Law
"Analyze class data to plan instruction. Hand Claude a folder of data — roster, diagnostics, attendance, your own notes — and it builds a clear picture of where every student is, allowing teachers to better tailor instruction. You control what data is shared; nothing you share is used in model training."
That marketing pitch speaks directly to overburdened teachers, but it risks encouraging them to bypass school- or district-level procurement and privacy safeguards. Public K–12 systems operate under specific legal frameworks — notably FERPA (the Family Educational Rights and Privacy Act) — that restrict how student data may be shared and with whom.
Two Critical Legal Conflicts
First, CfT’s terms require a teacher to be able to designate Anthropic as a "school official." Under FERPA, a vendor counts as a school official only when the district formally designates the vendor and exercises "direct control" over its use of data. An individual teacher generally lacks the authority to make that designation.
Second, the service's contract requires teachers to indemnify Anthropic for claims arising from uploaded data and caps the company’s liability. In effect, the contract shifts legal and financial risk from Anthropic to the individual educator — a dangerous position for teachers who may unknowingly upload sensitive records.
Why The Stakes Are High
If teachers accept the terms and upload class rosters, diagnostics, individualized education program (IEP) details, Section 504 plans, or medical information, they could violate district policy or federal law and face disciplinary or employment consequences. This is especially acute for students with disabilities or medical accommodations, where privacy protections are strongest and missteps carry serious legal risks.
A Better Path: District-Controlled Data Governance
We want tools like CfT to work and to be widely adopted — but adoption must occur through systems that districts control. State- and district-led initiatives such as the Texas Education Exchange, the Michigan Data Hub and the Ed Tech Collaboratory are developing standards-based, local-control approaches that allow AI models to access permitted data safely and compliantly. These models help preserve teacher time and improve instruction while protecting districts, educators and students from legal exposure.
What Anthropic And Other AI Labs Should Do
Building secure, governance-driven platforms is difficult and slower than a "move fast" product launch — but that deliberate approach is needed in public education. AI labs should partner with states and districts to design, pilot and fund interoperable, auditable data-access systems. Doing so would make these tools part of the solution rather than accelerating distrust and backlash against AI in schools.
Bottom Line
Claude for Teachers presents meaningful instructional potential. But its current marketing and contract terms create conflicting incentives that could put individual educators at legal and employment risk. Responsible deployment requires district-led governance, clear vendor designations under FERPA, and contracts that do not transfer unacceptable liability to teachers.
Help us improve.
