Most lists of AI tools for dyslexia are lists of things that read to the student. Text-to-speech, audiobooks, a browser extension that turns the page into a voice. Those are accommodations, good ones, and your school almost certainly has them already. They are not the interesting part of what AI can now do for a dyslexic reader.
The interesting part is a small group of tools that lower the decoding barrier without lowering the thinking, and a smaller group that help a student get better at reading rather than around it. This post is for special education teachers, IEP teams, reading specialists and the classroom teachers who have three students with dyslexia in a class of 28 and no extra adult. It is written for US schools first, with the FERPA, COPPA and hosting data you need before anything touches a student account, and it also serves UK and Australian readers who search for the same tools under SEND.
How This List Was Built
The pool was the 25 tools in the directory's special education and accessibility category, plus every tool in the wider directory of 90 AI tools for education that lists a dyslexia-friendly mode or a read-aloud feature. I looked for four things: does it help a dyslexic student read, write or show understanding in a way a plain chatbot cannot; is there evidence of classroom use; is the safety record clear enough to put in front of a district privacy officer; and does it keep the thinking with the student rather than doing the assignment for them. Nobody paid to be here and vendors cannot buy a place.
I left out four tools you will see recommended elsewhere. SchoolAI lists a dyslexia-friendly mode and its monitored tutors are good, but it is a general AI tutoring platform rather than a reading tool, and its own directory record declares that student data is shared with third parties, which is a conversation for your privacy officer, not a ranking. Wayground has a dyslexia font and read-aloud on quizzes, but it is a quiz platform and its record is blank on FERPA, COPPA and hosting. Glint builds visual supports and social stories, which is a different job. And MagicSchool's text leveller is teacher-facing and already covered in my MagicSchool alternatives post. Microsoft Immersive Reader, Read&Write and Speechify are not in the directory because they are assistive software rather than AI-first education products; use Immersive Reader anyway, it is free and already in your stack.
Pricing and features verified 5 September 2026.
What the Directory Numbers Say About Accessibility
Of the 90 tools in the aieducator.tools directory, only 10 list a dyslexia-friendly mode, and 26 list no accessibility features at all. That is the market talking: accessibility is still an afterthought for most education AI vendors. Within the special education and accessibility category, 14 of 25 tools declare FERPA compliance, 9 have no minimum age and so can be set up for elementary students, and just 3 carry an independent 9ine security assessment. Every one of those figures is a vendor declaration except the 9ine column, which is the independent one. A blank in our record means the vendor has not told us, not that the tool fails.
For scale on the need: the International Dyslexia Association estimates that as many as 15 to 20 percent of the population have some symptoms of dyslexia, and around 85 percent of students with a specific learning disability have a primary difficulty with reading and language processing. The National Center for Education Statistics counted 7.5 million students, 15 percent of all US public school students, served under IDEA in 2022-23, with specific learning disabilities the largest category at 32 percent. Dyslexia is not an edge case. It is the biggest single group in your special education caseload.
The Six Tools at a Glance
| Tool | Best for | Pricing | Grades | Min age | Privacy (declared) | Data hosted | Independent security assessment |
|---|---|---|---|---|---|---|---|
| Diffit | Levelled texts for the whole class | Freemium | K-12 | Not stated | FERPA, COPPA (vendor site) | Not stated | No |
| Kami | Read-aloud and voice inside worksheets | Freemium | All | No minimum | FERPA, COPPA, GDPR | US | Yes (9ine) |
| Readling | Oral reading fluency, elementary | Paid | K-8 | No minimum | GDPR | UK | No |
| EchoGuide Pro | Private reading rehearsal | Paid | All | 4+ | FERPA, COPPA, GDPR | UK | No |
| Knowt | Notes and revision, dyslexia mode | Freemium | All | No minimum | FERPA, COPPA, GDPR | US | Yes (9ine) |
| Snorkl | Showing understanding by voice | Freemium | K-12 | Not stated | FERPA, COPPA (privacy policy) | Not stated | Yes (9ine) |
The privacy column is what each vendor declares, not an audit. The final column is the independent one.
Best for Levelled Texts: Diffit
Diffit is the fastest way to give a dyslexic student the same article as everyone else at a reading level they can decode. You paste a URL, a PDF, a YouTube link or a topic, pick a grade level, and it produces an adapted text with vocabulary, comprehension questions and a summary, in about a minute.
Teachers in our community forum name Diffit more than any other tool for inclusion, specifically for generating levelled texts and comprehension questions, and the principle they landed on in that thread is the one I would put on the wall: the best tools reduce barriers without removing challenge. Against MagicSchool's text leveller, Diffit wins on output quality and on the fact that the whole resource, not just the passage, is rebuilt at the new level.
Watch out for the free tier, which only keeps content from the last 90 days and exports to PDF and print, not Google Classroom. And levelling a text is a teacher decision, every time: if you drop a grade 8 reader to grade 3 vocabulary, you have removed the challenge along with the barrier. Best for whole-class differentiation where one student needs a lower decoding load. Not for building the student's own reading skill; it does none of that.
Pricing: Basic is free. Premium is $14.99 per month or $149.99 per year. Schools pay a flat annual rate tiered by enrolment, quote-based (verified 5 September 2026).
Safety: our directory record is blank on FERPA and COPPA, so I checked. Diffit's privacy page states that it is FERPA and COPPA compliant, collects no student data, has signed the National Data Privacy Agreement and holds a Common Sense Media pass. It does not disclose where data is hosted. No independent 9ine assessment.
Best for Reading and Writing Inside the Worksheet: Kami
Kami is the tool that turns the PDF you were going to hand out into something a dyslexic student can hear, annotate and answer by voice. It sits over any document in Google Drive or OneDrive. The Read Aloud tool speaks the page with adjustable speed, Voice Typing lets the student dictate into a text box, and Voice Comments let them record an answer instead of writing it. Kami's own special education guide describes exactly this workflow.
What sets it apart from Immersive Reader is that the student does not leave the assignment. The reading support, the answer and the teacher's feedback live in the same file, which is what makes it survive contact with a real Tuesday. Kami also has the cleanest safety record on this list: FERPA, COPPA and GDPR declared, data hosted in the US on AWS and Google Cloud, no minimum age, and an independent 9ine assessment.
Watch out for two things. Kami is an all-purpose classroom platform, so the accessibility tools are one corner of a big product and staff need showing where they are. And the read-aloud is a synthetic voice reading whatever is on the page; if the worksheet is badly written, it reads badly. Our directory record does not yet list Kami's accessibility features; that is a gap in our data, not in the product. Best for any school already on Google or Microsoft that wants one accommodation layer over everything. Not for reading instruction.
Pricing: Basic is free with ten or so core tools. The Teacher plan is $149 per year. School and district plans are custom quotes (verified 5 September 2026).
Safety: FERPA and COPPA declared, no minimum age, hosted in the US, 9ine assessed, GDPR declared. Collects no student data by its own declaration.
Best for Oral Reading Fluency in Elementary: Readling
Readling is the only tool here that helps a young student get better at reading rather than around it. The child reads aloud from any physical book, the app listens, and the AI gives feedback on fluency and expression, with pronunciation guidance and quick comprehension quizzes. Teachers get a dashboard of who read, for how long and where they stumbled. Parents get a QR code so reading at home counts too. It is built for UK primary schools, which US readers should hear as K-5 and early middle school.
The reason it makes this list rather than the many phonics apps that do not is the dyslexia-friendly font and the fact that it works from the books already in your classroom. The obvious alternative is Google's Read Along, which is free; Readling's edge is the teacher dashboard and the parent loop. Watch out for the UK origin: the interface is British English, the curriculum references are English, and there is no FERPA declaration because FERPA does not apply to a UK company, so a US district will need its own data agreement. Best for elementary reading intervention and daily reading practice. Not for high school.
Pricing: Paid, per school. There is no public pricing on the vendor site at the time of checking; expect a quote (verified 5 September 2026).
Safety: FERPA and COPPA marked not applicable (UK vendor). No minimum age. Collects a student's name and learning data, does not share with third parties, deletes on request. Hosted in the UK. GDPR declared. No independent 9ine assessment.
Best for Private Reading Rehearsal: EchoGuide Pro
EchoGuide Pro gives a dyslexic student somewhere to practise reading aloud where nobody is listening. It highlights the next word as the student reads, guides pace, flags clarity and gives fluency feedback in real time, with a dyslexia-friendly mode and a low-anxiety design. Anyone who has watched a twelve-year-old with dyslexia made to read a paragraph to the class knows why that matters.
Against Readling, EchoGuide is the older student's tool: it runs from age 4 to adult, it covers speaking and presentation practice as well as reading, and it is usable in history and geography lessons, not just English. Watch out for its age: the company launched in 2026, so the classroom evidence is thin and I would run it as a pilot, not a purchase. It also has no independent security assessment yet. Best for middle and high school students rehearsing reading and speaking on their own. Not for structured phonics intervention.
Pricing: Paid, with a seven-day free trial. No public pricing on the vendor site at the time of checking (verified 5 September 2026).
Safety: FERPA and COPPA declared, minimum age 4, collects no student data by its own declaration, hosted in the UK, GDPR declared. No 9ine assessment.
Best for Notes and Revision: Knowt
Knowt is the study platform I would hand to a dyslexic high school student, because it has a dyslexia-friendly mode, a screen-reader compatible interface, adjustable text and high-contrast options, and it turns whatever they upload into notes, flashcards and practice questions they can hear as well as read. Students can also record a lesson and get notes from it, and that is the feature that changes a dyslexic student's week, because the bottleneck was never understanding the lesson, it was writing fast enough to keep it.
Against Quizlet, Knowt wins on the accessibility settings and on the safety record: FERPA, COPPA and GDPR declared, US hosting, no minimum age, a free district DPA and an independent 9ine assessment. Watch out for the fact that AI-generated notes are a summary of what the recording caught, so students still need to check them, and for the free tier's AI limits, which bite quickly. Best for grades 6 to 12 students who need to get information out of their head and onto a page. Not for early readers.
Pricing: Basic is free. Ultra is $24.99 per month, or $149.99 per year (billed as $12.49 per month). School and district licences from $2.75 per student, with data privacy agreements signed free (verified 5 September 2026).
Safety: FERPA and COPPA declared, no minimum age, hosted in the US, 9ine assessed, GDPR declared. Collects student data, does not share it with third parties, deletes on request.
Best for Showing Understanding by Voice: Snorkl
Snorkl lets a dyslexic student prove they understand something without writing a paragraph. The student talks through their thinking while drawing or working on a whiteboard, and the AI gives feedback on the reasoning against criteria the teacher set. For a student who can explain the water cycle and cannot spell precipitation, that is the difference between a grade that measures science and one that measures dyslexia.
I covered Snorkl at length in my post on AI-resistant assessment tools, and it earns a place here for a different reason: it is the cleanest example of assessing the process and the performance, not just the written product. Against SayVeritas, Snorkl is free for teachers, works from kindergarten up and is built around drawing plus voice rather than oral exams. Watch out for the accent handling with the youngest and most anxious speakers, which needs a trial run, and for the free tier's activity cap. Best for maths and science explanations in K-12. Not for extended writing.
Pricing: The teacher plan is free with a limited number of activities. School and district plans are quote-based (verified 5 September 2026).
Safety: our record is blank on FERPA and COPPA, so I read the privacy policy. It addresses FERPA-covered data directly and puts COPPA consent for under-13s with the school, which is the standard model. Hosting location is not disclosed. Snorkl does carry an independent 9ine assessment.
How to Actually Choose
Split the job in two before you look at any tool. Decoding is the doing; comprehension, argument and expression are the thinking. Accommodate the doing as hard as you like and protect the thinking at all costs.
Accommodate the decoding. Never the thinking.
Three rules follow from that. If a student cannot access the text, use Kami or Diffit and lower the decoding load, then set the same comprehension question everyone else gets. If a student cannot show what they know because the output is writing, use Snorkl or Kami's voice tools and assess the explanation. If the goal is for the student to read better, not just cope, use Readling in elementary or EchoGuide Pro later, and measure fluency over a term, not a lesson.
One more rule for the leadership team. Before any of these touches an IEP student, get the vendor's FERPA and COPPA position in writing, check the minimum age against the students who will use it, and ask where the data lives. The directory's security assessments page tells you which tools have had an independent look. Two of the six here have not, and one of those launched this year.
What I'd Do This Week
Pick one class and one dyslexic student in it. Put the next reading assignment through Kami, switch on Read Aloud and Voice Comments, and let the student answer by voice. Keep the question identical to everyone else's. At the end of the week, compare the quality of that answer with the last written one. If the thinking was always there and the writing was hiding it, you now have evidence for the IEP meeting, and a reason to look at the other five.
The Bottom Line
For a school that wants one accommodation layer, Kami; for levelled texts, Diffit; for elementary reading fluency, Readling; for private rehearsal in older grades, EchoGuide Pro; for study and notes, Knowt; and for showing understanding without writing, Snorkl. Only 10 of the 90 tools in the directory bother with a dyslexia mode, so start with the ones that did. Compare the rest in the special education and accessibility category, and if you are choosing tools for the wider SEND or special education team, my guide to AI tools for special education teachers and the post on AI tools for IEP writing pick up where this leaves off.
The student who cannot read the worksheet was never the problem. The worksheet was.
Dan Fitzpatrick is a Forbes contributor, bestselling author and founder of The AI Educator, and has trained more than 150,000 educators in over 30 countries. More from Dan.
