Which AI Teaching Tools Leave Privacy Questions Unanswered?

Of the 161 AI tools in the AI Educator Tools directory on October 22, 2026, 116 have privacy details recorded and 45 have none. The blanks cluster among the best-known tools: all seven listings with 20 or more reviews, including ChatGPT and Google Gemini, have no recorded privacy fields. Read a blank as "not yet recorded", then check the vendor's own privacy page before any student data goes in.

This is for the person at a school or district who has to say yes or no to a tool: a data protection lead, a technology director, or a principal handed a spreadsheet of requests. The figures come from our own directory, so they describe what we have recorded, not the whole market. Gallup's survey of 1,952 U.S. K-12 educators, fielded July 28 to August 14, 2026 and published October 14, found that about three in five used AI tools for work last school year (Gallup). With that many staff already using these tools, the privacy fields are what a school checks first, and a blank leaves it with nothing to check.

How was the data gathered, and what does a blank mean?

We read every listing from the directory's public API on October 22, 2026 (161 listings; Ask Poppy and Madlen each appear twice and are counted as listings, not deduplicated). For each we counted the fields a school asks about first: FERPA, COPPA and GDPR status, whether the tool collects student data, whether it shares data with third parties, data hosting location, minimum user age and whether an independent security assessment by 9ine is on file.

Three limits matter. Most fields are the vendor's own statement, recorded by us, not an audit. The exception is the 9ine assessment, which is independent (see what the assessments cover). And a blank is our gap, not a finding about the vendor: all 45 blank listings were added between February 9 and March 13, 2026, and every one of them is blank on every privacy field at once, which points to fields not recorded at the time, not to 45 separate vendor refusals. We run the directory, so treat the percentages as a picture of our own records and check the vendor before relying on any of them.

Which kinds of tool have the least privacy information recorded?

Presentation and design tools (half blank) and school administration tools (44%) have the thinnest records. The table counts listings by directory category.

Categories of seven to twelve listings move a lot on one or two tools, so read the small rows as a hint, not a ranking. The two large ones are steadier: assessment and grading is the best-documented big category at 20% blank, and lesson planning, at 29%, sits close to the overall 28%. If you are comparing administration tools specifically, the School Administration and Safety category is where to start.

Are the best-known tools the least documented?

Yes, on our records, and it is the most useful thing in the data. We sorted listings by review count. Of the seven with 20 or more reviews, all seven have no privacy fields recorded. Of the 14 with 10 or more, 12 are blank.

The likeliest reason is how the directory grew, not anything about these vendors, though this is our inference from the listing dates. The oldest listings have had the longest to collect reviews, and all 45 blanks were added in February or March 2026. The same pattern would explain a second result: 37 listings carry an independent 9ine assessment, and 32 of them have no recorded fields. The assessments exist, and our fields have not caught up. Only five assessed listings have both, among them Kami, Knowt, EnlightenAI and TeachAid.

What this means for a school: popularity and a green shield tell you a tool has been looked at, not that you can read its terms in one place. Do the reading yourself for any tool you approve, however well known.

What do the 116 documented listings say?

They say vendors mostly claim FERPA and GDPR alignment where they say anything, and that a sizeable minority collect student data. These are the figures for the 116 listings with fields recorded.

Two rows deserve a second look before you rely on them. First, 61 listings say they collect student data, and for 32 of those the FERPA field reads "not applicable". There can be good reasons, such as a tool used outside the United States, but when a tool takes student data and the FERPA box says not applicable, ask the vendor why. Second, ten listings record that they share data with third parties. That is a field to read in full on the vendor's page, not a verdict.

For the law behind the columns: the Federal Trade Commission says the COPPA Rule was "designed to protect children under age 13" and applies to operators of services directed to children under 13, or that know they are collecting personal information from them (FTC COPPA FAQ, page modified July 10, 2026). The U.S. Department of Education describes FERPA as giving parents the right to access their children's education records, to seek amendments, and to control disclosure of personally identifiable information, with those rights transferring to the student at 18 or on entering a postsecondary institution (Student Privacy Policy Office).

Where is the data hosted, and what are the minimum ages?

Most recorded hosting is in the United States, with a substantial European and UK share. Of the 110 listings that give a location, 53 name the United States, 24 the European Union (several naming Germany or writing "European Union"), 19 the United Kingdom, and 14 give another country, several regions, or a note that the location is unverified.

Minimum user age, for the 116 listings that record it, is spread widely: 37 say 13+, 33 say 18+, 24 set no minimum, 11 say 8+, five say 4+, four say 16+ and two say 6+. A tool with no stated minimum age is not a tool approved for young children. It is a tool that has not said, and COPPA's under-13 line is where that silence becomes your problem as the adult deploying it.

What should you check when a listing is blank?

Do a five-minute check from the vendor's own pages, and write the answers down. This is the checklist we would use, built from the fields above. Copy it into your approval spreadsheet.

Score each row "found", "not found" or "red flag". Approve nothing with a red flag or with more than two "not found" rows, and re-run the check when the vendor changes its terms. For a wider approval process, our guides on which AI tools a school can approve and AI tools for district technology directors walk through the steps, and FERPA-compliant AI tools for teachers covers that single field in depth.

How should a school treat the blanks it finds in directories like ours?

Treat them as a work list for you and for us. For you, a blank means the check above is the whole job, not a shortcut: open the vendor pages, fill the table and keep the answer with the approval. For us, it means more records to complete, and a count like this one is only as good as the records behind it.

Our own view, and it is only that, is that a school should never approve a tool on a directory entry alone, including ours. The directory is a place to find candidates and to see whether someone independent has looked. The decision rests on the vendor's current terms and your own data agreement.

Last checked October 22, 2026. All counts come from the directory's public API on that date. Change log: first edition.