AI Tutors and the Data They Collect
A plain-language guide to ai tutors and the data they collect: the record, the purpose, the authority and the safeguard.
- Record
- student input to an AI tool in ai tools and student privacy
- Data
- AI Tutors and the Data They Collect
- Safeguard
- a pre-adoption review and a contract term on training use
- First check
- State the purpose in a sentence a student could understand.
The short answer
For ai tutors and the information they collect, think in terms of a chain: a student, a student input to an AI tool, a purpose, an authority, a recipient, a safeguard and a storage date. Break any link and the sharing fails, even if every other link is sound. This guide walks that chain link by link, so the reader can see exactly where a proposed sharing holds and where it breaks.
A breach response that starts with notification and skips containment shall notify students about a problem that is still getting worse.
Support staff access to live files is a sharing in substance, even when the vendor describes it as maintenance.
A notification that no student can find, or that no student can understand, does not do the work of consent or of a directory designation.
Logs turn a correct determination into a defensible one, because a reviewer, an auditor or a student will eventually request what happened and why.
What this covers
Accessibility data in an education platform can reveal disability status, and access to it ought to be limited to employees who need it for support.
Former students retain rights over files retained regarding them, and alumni processes should not treat those files as unprotected history.
Counselling records raise the same boundary question, and employees ought to check the exemption before treating such a record as a routine file.
Law enforcement records created by a site police unit for a law enforcement purpose are treated separately from education records, with their own limits.
Voice information in support and proctoring platforms is biometric in substance, even when vendors describe it as audio processing.
Assistive technology can collect usage information that reveals disability, and vendors of such tools ought to be reviewed with that sensitivity in mind.
How to work through it
Answer complaints about access in writing, with reasons, because an unexplained refusal cannot be challenged or learned from.
Retain inventories current, and treat an inventory that is a year out of date as no inventory at all.
Review shared folders periodically, and treat shared-folder review as a confidentiality task with an owner, not as housekeeping.
Assess AI tools with the same gate, plus a particular review of training information use, prompt retention and output handling.
Require a court order or subpoena procedure for compelled disclosures, and do not disclose on the strength of a letter alone.
Assess biometric platforms with the highest gate of all, since the data cannot be reissued if compromised.
Check consent for specificity, freedom and revocability, and do not rely on a blanket consent that was never explained.
Review a determination after it is made: was the information used for the stated objective, does the receiver still need it, has retention expired?
A worked illustration
In that illustration the logs are released solely to the extent the procedure needs, with redactions for students not involved.
Consider an employer who asks an institution to confirm a graduate's degree and is given solely the directory facts the institution has designated, not the transcript.
Consider an admissions department that retains applicant information for students who never enrolled, and that deletes those records when the specified schedule expires.
The cost of fixing the badly done adoption, in renegotiation, migration and lost trust, is far higher than the cost of the review that was skipped.
In that illustration the answer depends on whether the notes were ever shared, and the service records the reasoning for future cases.
Where guides go wrong
This guide does not track every state rule or agency update in real time, and readers ought to verify current authoritative advice before acting on a state-particular issue.
Accepting a checkbox as verifiable parental consent is an error, since a checkbox that anyone may tick verifies nothing.
Treating a vendor as if it owned the data it processes is a governance failure, even when the vendor's sales material encourages that view.
Leaving access in place following a role change is a mistake that compounds quietly, because the former access looks legitimate in every log.
Answering a complaint about denied access without reasons is an error, because an unexplained refusal can not be challenged or learned from.
Sources and evidence on the record
- Family Educational Rights and Privacy Act regulations, 34 CFR Part 99, text on eCFR.
- US Department of Education, Student Privacy Policy Office, studentprivacy.ed.gov - FERPA guidance, FAQs and model notices.
- US Department of Education, ed.gov - guidance, Dear Colleague letters and enforcement information.
Evidence basis: 34 CFR Part 99, Department of Education guidance, FTC AI guidance Applies when: When a college holds, uses or discloses a student input to an AI tool in the context of ai tools and student privacy