Get the weekly research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean. No account.
The policy discloses that Khan Academy's AI-powered features for schools, including Khanmigo and Writing Coach, are covered by service agreements prohibiting AI model providers from using student input data to train their models. Users are also advised not to submit personal information in inputs to AI features.
This analysis describes what Khan Academy's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
This provision establishes a contractual restriction on AI model providers regarding student input data, which is operationally significant for FERPA compliance and applicable U.S. state student data privacy laws. Compliance teams should verify that the underlying vendor agreements with AI providers include enforceable, auditable restrictions consistent with this disclosure.
Interpretive note: The policy asserts that vendor agreements prohibit training-data use but does not disclose the specific terms, audit rights, or enforcement mechanisms in those agreements, making independent verification of compliance uncertain.
Under this provision, the agreement states that AI model providers powering Khan Academy's school-facing AI tools are contractually prohibited from using student input data to train their models. Students and school administrators are advised by the policy not to include personally identifiable information in AI feature inputs.
Cross-platform context
See how other platforms handle AI-Enabled Features and Training Data Restriction and similar clauses.
Compare across platforms →Monitoring
Khan Academy has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 20 platforms.
"Our services provided to Schools may include use of AI-Enabled Features such as our AI-powered teacher tools, our student tutor chatbot, Khanmigo, and Writing Coach features. AI-Enabled Features included in School services are designed to comply with applicable U.S. student data privacy laws. We provide privacy guidance to our users informing them that AI-Enabled Features are not intended to be used to process personally identifiable information, and advising them to not include personal information in inputs submitted to AI-Enabled Features. To provide further privacy assurances, our service agreements with the providers of artificial intelligence models used to power our AI-Enabled Features do not permit the providers to use Input data to train their models.Excerpt from Khan Academy's Privacy Policy
1. REGULATORY LANDSCAPE: This provision engages FERPA, which governs the disclosure and use of student education records, and COPPA for school users under 13. Applicable U.S. state student data privacy laws (including California's SOPIPA and equivalent statutes in other states) may impose additional restrictions on the use of student data by third-party AI providers. The U.S. Department of Education and applicable state education agencies hold enforcement authority. The FTC may also be relevant if AI data practices are found to be deceptive relative to disclosed commitments. 2. GOVERNANCE EXPOSURE: Medium. The policy asserts that vendor agreements prohibit training-data use, but does not disclose whether these agreements include audit rights, incident notification obligations, or independent verification mechanisms. The adequacy of this contractual commitment as a compliance control depends on the enforceability and specificity of the underlying DPA language, which is not reproduced in the policy. 3. JURISDICTION FLAGS: California SOPIPA explicitly prohibits operators of online services directed to K-12 students from using covered information for non-educational purposes, including model training. Other states with student data privacy laws (New York, Texas, Illinois) impose similar restrictions. EU/EEA school users would require evaluation under GDPR's requirements for data processor agreements and lawful basis for AI processing. 4. CONTRACT AND VENDOR IMPLICATIONS: School procurement teams and district legal counsel should request copies of Khan Academy's DPAs with AI model providers to verify that the prohibition on training-data use is contractually specific, enforceable, and includes breach notification and audit provisions. The policy's characterization of AI features as not intended to process personally identifiable information does not constitute a technical guarantee that such processing cannot occur if students submit personal information in inputs. 5. COMPLIANCE CONSIDERATIONS: Districts integrating Khan Academy's AI features should document their review of the stated DPA commitments as part of vendor assessment procedures under applicable student privacy laws. Schools should establish internal guidance reminding students not to submit personally identifiable information in AI feature inputs, consistent with Khan Academy's stated advisory.
Regulatory citations, enforcement risk, and due diligence action items.
Provision-level monitoring, governance timelines, and regulatory mapping built from archived source documents and historical version tracking.
This provision establishes a contractual restriction on AI model providers regarding student input data, which is operationally significant for FERPA compliance and applicable U.S. state student data privacy laws. Compliance teams should verify that the underlying vendor agreements with AI providers include enforceable, auditable restrictions consistent with this disclosure.
Under this provision, the agreement states that AI model providers powering Khan Academy's school-facing AI tools are contractually prohibited from using student input data to train their models. Students and school administrators are advised by the policy not to include personally identifiable information in AI feature inputs.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Khan Academy.