TeacherAssist.ai
For district & building administrators · updated August 2026

Build a clear AI policy, and give your teachers a tool that's built to follow it.

As AI-in-education legislation takes shape, districts have a real opportunity to get ahead of it — with a clear policy and a tool built around it from day one. TeacherAssist.ai takes a fundamentally different approach: the AI doesn't grade. It evaluates a student's handwritten work and generates the standards-aligned data a teacher needs to dig in and grade it themselves — the same granular insight teachers expect from digital assignments, now available from pen and paper.

What's probably already happening

Shadow AI grading has no controls, no paper trail, and no deletion policy

A teacher trying to save time on grading doesn't need district approval to open ChatGPT, Gemini, or Claude in a browser tab and paste in a stack of student essays. It's fast, it's free, and it's completely outside your visibility. Here's what that actually means.

No data agreement

Student work uploaded to a personal consumer AI account, with no FERPA-oriented agreement covering how it's used or stored.

No deletion guarantee

That student work can sit in a personal chat history indefinitely, with no district control over how long it's retained.

No enforced human review

Nothing technical stops a teacher from copying whatever score the AI hands back straight into the gradebook.

No administrator visibility

No audit trail, no policy, no way to know it's happening in your own building until something goes wrong.

TeacherAssist.ai exists to close that gap — without asking AI to replace the teacher's judgment. It evaluates the handwritten work and generates the data a teacher needs to grade it themselves, backed by the data agreement, deletion policy, and review requirement a browser tab full of ChatGPT will never have.

Why this is about to matter more

AI-in-education law and guidance, plain English

Most states aren't passing binding AI-in-education laws — they're publishing guidance documents instead: non-mandatory recommendations that leave the actual decisions to districts. This page tracks both the Illinois legislation and the state guidance most relevant to how teachers use AI to grade or generate curriculum, and is updated as either one changes.

Enacted = binding law  ·  Guidance = a published, non-mandatory recommendations document  ·  Dead = did not pass

SB 1920 (2025)Illinois · Enacted

Statewide K-12 AI guidance mandate

Passed the House 74–34 with unanimous Senate support in spring 2025. Directs the Illinois State Board of Education to publish statewide K-12 AI guidance — covering data privacy, bias, AI literacy, and classroom use — by July 1, 2026. The law itself doesn't set any AI-use rules; it only required ISBE to write the guidance below.

In effect
ISBE AI GuidanceIllinois · Guidance

409-page K-12 AI guidance document

Published July 9, 2026 to fulfill the SB 1920 mandate above. Non-mandatory recommendations, not a compliance requirement — districts decide how to apply it locally. Its core framing: AI should inform teaching and preserve human relationships in the classroom, not replace the teacher's judgment. That's the same principle behind TeacherAssist.ai's AI-evaluates, teacher-grades workflow. It also recommends districts pilot new AI tools with a small group before rolling them out widely, and advises teachers to include state learning standards directly in their prompts so AI output stays anchored to grade-level expectations.

Find it on ISBE.net (search "AI Guidance") ↗
Non-mandatory
PA 104-0201Illinois · Enacted

Community college AI instruction ban

Signed August 2025, effective January 1, 2026. Bars community colleges from using AI as the sole source of instruction for a course. A separate law from SB 1920 above; it doesn't apply to K-12 grading.

In effect
SB 416Illinois · Did not pass

Student Educational Technology Rights Act

Passed the Senate 57–0, but stalled in the House and was re-referred to the Rules Committee on May 31, 2026 — dead for this session. Would have made human-in-the-loop requirements for student-data AI tools binding law rather than guidance; TeacherAssist.ai's AI-evaluates, teacher-grades workflow already matches what it proposed.

Did not pass
HB 1519Missouri · Tracking

AI in Education Task Force

Would direct the state board of education to convene a task force studying AI use in K-12 classrooms, including grading, and to issue guidance to districts. Still in committee — separate from the guidance DESE has already published below.

In committee
DESE AI GuidanceMissouri · Guidance

AI Guidance for Local Education Agencies, v1.0

Released by the Missouri Dept. of Elementary and Secondary Education for the 2025-26 school year, developed with Missouri's Computer Science Advisory Council. Non-mandatory; emphasizes human oversight and AI as a tool to enhance, not replace, educators — consistent with TeacherAssist.ai's architecture.

View the guidance (PDF) ↗
Non-mandatory
CA Model AI PolicyCalifornia · Guidance

CDE Model Policy: AI in Education

Required by SB 1288 (2024); released by the California Dept. of Education in 2026 as a 26-principle model policy districts may adopt or adapt. Non-mandatory under Ed Code 33308.5. Explicitly states the educator of record retains sole authority to determine final grades, regardless of how AI is used in the process — the most direct match to TeacherAssist.ai's architecture of any state guidance we track.

View on CDE.ca.gov ↗
Non-mandatory
NYC AI PlaybookNew York · Guidance

NYC Public Schools Guidance on AI

New York State has no statewide K-12 AI guidance yet (a bill to create one, A6972, is still pending). New York City Public Schools — the nation's largest district — released its own preliminary AI guidance in March 2026, explicitly barring AI from assigning grades or discipline while inviting community feedback ahead of a fuller playbook.

View NYC Public Schools guidance ↗
District-level
No statewide guidanceTexas · Tracking

Texas has not yet issued state guidance

The Texas Education Agency has not published statewide K-12 AI guidance as of this writing. Activity so far has come from outside philanthropy — the Communities Foundation of Texas launched an AI + Education Leadership Collaborative to help districts navigate policy in the meantime.

No guidance yet
National viewAll 50 states · Tracking

Where the rest of the country stands

At least 34 states have now issued some form of K-12 AI guidance, and a smaller number have passed binding legislation — most states are choosing non-mandatory guidance over law, the same path Illinois and California have both taken. The requirement that keeps reappearing across nearly all of them, guidance or law: a teacher, not the algorithm, assigns the final grade.

Ongoing
134
AI-in-education bills introduced across 31 states in 2026
~12
states with binding K-12 AI legislation enacted, not just guidance
~19
States with AI-in-education bills still pending
34+
states that have issued K-12 AI guidance rather than binding law

Two states, the same non-mandatory approach

How Illinois's guidance compares to California's

Illinois and California have both landed on the same strategy: publish detailed, non-mandatory guidance rather than pass binding law. Here's how the two documents actually compare.

Illinois — ISBE Guidance
California — CDE Model Policy
Legal basis
SB 1920 (2025) directed ISBE to write it
SB 1288 (2024) directed CDE to write it
Released
July 9, 2026
2026
Mandatory?
No — non-mandatory recommendations
No — Ed Code 33308.5 states compliance "is not mandatory"
Format
409-page narrative guidance document
26-principle model policy, adoptable as a template
Grading language
General principle: AI should inform teaching, not replace it
Explicit: the educator of record retains sole authority to determine final grades
Data privacy
Flags student data privacy as a key ethical issue for districts
Explicit ban on using student data to train AI models or for advertising
Vendor evaluation
Recommends piloting AI tools with a small group before district-wide rollout
Calls for pre-deployment evaluation and enforceable vendor safeguards
Standards in prompts
Recommends teachers include state learning standards directly in AI prompts, so output is anchored to grade-level expectations
Not directly addressed
The bottom line for districts: neither state has passed binding law on AI grading yet, but both point the same direction — human oversight, teacher-held grading authority, and data-privacy safeguards. California's policy simply says it more explicitly. TeacherAssist.ai's AI-evaluates, teacher-grades architecture was designed to align with both.

Requirement by requirement

How TeacherAssist.ai is built to align

Across nearly every state bill and guidance document we track, the same handful of requirements keep reappearing. Here's how our platform was already designed around each one — before any of these bills or guidance documents existed.

Keep AI output grounded in grade-level standardsIllinois's guidance recommends teachers include state learning standards directly in AI prompts so output stays anchored to grade-level expectations.
Standards baked into every evaluation, not typed inTeacherAssist.ai reads each teacher's active state standards directly from their profile and builds them into every grading prompt automatically — no manual prompt-writing required.
The teacher assigns every gradeAI evaluation never becomes a final grade without the teacher reviewing the actual student work.
AI evaluates, teacher gradesEvery batch evaluation job returns diagnostic data to the teacher's queue; the teacher reviews the work and enters the grade — the AI never does.
No AI-generated score to rubber-stampSeveral proposals raise concern about a teacher being tempted to copy an AI-generated number straight into the gradebook without actually reviewing the work.
No grade-ready number existsTeacherAssist.ai's AI never outputs a score — only an achievement level and diagnostic data. The teacher still has to look at the work to assign a grade. See below.
Minimizing student personal dataReducing the personally identifiable information collected and retained on students.
Initials-only, delete-after-evaluationFull names, IDs, and biometric data are never collected; uploaded work is deleted once the AI evaluation completes.
Disclosure of which AI models are usedDistricts and boards increasingly want to know exactly which AI systems touch student work.
AI Model Disclosure SheetA one-page document naming every model in the pipeline, ready for board-approval packets.
No use of student data to train AI modelsPreventing vendors from using student work to improve their own or third-party AI systems.
Contractual model protectionsEnterprise data-processing terms with our AI providers prohibit training on customer data.
A written data processing agreementA signed record of how student data is handled, retained, and deleted.
District DPA on requestA FERPA-oriented Data Processing Agreement available for district counsel to review before signing.

What's actually unique here

AI evaluates the handwriting. The teacher still does the grading.

TeacherAssist.ai's AI doesn't grade — it evaluates. It reads a student's actual handwritten work, checks it against the standard, and generates the kind of granular, standards-level data that has typically only been available from typed or digital assignments. That data was never available from a stack of paper before. Now it is — without asking a single student to trade a pencil for a keyboard. The teacher still reviews the work and assigns the grade; the AI just gives them somewhere to start digging.

ExemplaryExceeds the standard
ProficientMeets the standard
DevelopingApproaching the standard
EmergingEarly progress
Needs SupportRequires intervention

Nothing is lost — it's redirected toward instruction. Behind every achievement level, TeacherAssist.ai generates the full granular, standards-aligned evaluation data. The teacher uses the badge to see at a glance who's excelling or struggling, then dives into that underlying data and the actual student work to plan differentiated instruction — small groups, reteach targets, enrichment — and to assign the grade.

Built for pen and paper, not against it. Because the AI evaluates handwriting directly, teachers don't have to move an assessment onto a screen to get this level of insight. TeacherAssist.ai makes it possible to keep assigning handwritten work without giving up the data that used to require digital tools.

📋

This is a teacher's assistant. Achievement levels and the evaluation data behind them live inside the teacher's TeacherAssist.ai workspace. It's an assistant that helps a teacher evaluate handwritten work and plan instruction, not a portal, a grading engine, or a report card replacement.

Under the hood

The compliance architecture, in detail

Data minimization

Student initials only. No full names, IDs, photos, or demographic data are collected at any point.

Automatic deletion

Uploaded student work and OCR text are deleted immediately once the AI evaluation is complete.

AI evaluates, teacher grades

AI generates diagnostic, standards-aligned data from the student's handwritten work — never a grade. The teacher reviews the actual work and enters every grade.

Encrypted infrastructure

Data encrypted in transit and at rest on Microsoft Azure, with authenticated access throughout.

No model training on student data

Enterprise agreements with our AI providers contractually prohibit training on customer data.

District-level DPA

A Data Processing Agreement naming subprocessors, retention periods, and breach-notification terms.

Frequently asked

Common questions from districts

Are teachers already using ChatGPT or other consumer AI to grade student work?

Anecdotally and increasingly in surveys, yes — many teachers use consumer AI tools like ChatGPT, Gemini, or Claude to grade papers for them, often without district knowledge or any data processing agreement in place. That exposes districts to the same FERPA and data-privacy questions pending legislation is trying to address, with none of the human-review or data-deletion controls a purpose-built classroom tool provides.

Is AI allowed to grade student work in Illinois?

Yes. Illinois has not banned AI-assisted grading. SB 1920 (2025) directed ISBE to publish statewide K-12 AI guidance, which it did on July 9, 2026 — a 409-page, non-mandatory document, not a legal requirement. SB416, which would have made human-in-the-loop requirements binding law, passed the Senate 57–0 but stalled in the House and died in the Rules Committee in May 2026.

What does "human-in-the-loop" AI grading mean, and how is TeacherAssist.ai different?

Most "human-in-the-loop" models mean an AI generates a suggested score that a teacher can override. TeacherAssist.ai goes a step further: its AI does not generate a score or a grade at all. It evaluates a student's handwritten work against the standard and generates diagnostic data, which the teacher then uses to review the work and assign the actual grade.

Does TeacherAssist.ai store student names or personal data?

No. Student work is identified by initials only, and uploaded files are deleted from storage once the AI evaluation is complete. Full names, ID numbers, and biometric data are never collected.

Is TeacherAssist.ai FERPA compliant?

TeacherAssist.ai is designed to support FERPA compliance and operates under a Data Processing Agreement with each district, acting as a school official with a legitimate educational interest under 34 CFR 99.31(a)(1). We recommend district counsel review the DPA before signing — how a tool is deployed by the district also affects FERPA compliance.

Why does TeacherAssist.ai show an achievement level instead of a score?

TeacherAssist.ai's AI doesn't generate a grade at all — it evaluates a student's handwritten work against the standard and generates diagnostic data. That data surfaces as an achievement level — Exemplary, Proficient, Developing, Emerging, or Needs Support — so a teacher can see at a glance where a student stands, then dig into the actual work and the underlying data to assign the grade themselves.

Does TeacherAssist.ai require students to type their work instead of writing by hand?

No — it's built for the opposite. TeacherAssist.ai evaluates a student's actual handwritten work and generates the same granular, standards-level data that has typically only been available from typed or digital assignments. That means teachers can keep assigning pen-and-paper work without giving up the diagnostic data digital tools provide.

Who sees the achievement badges?

Achievement badges are a teacher-facing tool inside the TeacherAssist.ai workspace, used to highlight where a teacher should focus. TeacherAssist.ai is an assistant to the teacher — not a portal or a report.

Does ISBE's new AI guidance require any changes to how we use TeacherAssist.ai?

No. ISBE's July 2026 guidance is non-mandatory and leaves specific policy decisions to each district, but its central theme — AI should inform teaching, not replace the teacher's judgment — is the same principle TeacherAssist.ai's AI-evaluates, teacher-grades workflow was already built around. We'll update this page if that changes, or if SB416-style legislation is reintroduced and passes in a future session.

Not legal advice. This page summarizes publicly available legislative and state-guidance information as of August 2026 for general awareness and is not a legal opinion. Legislation can change quickly and guidance documents are non-binding; districts should confirm current bill status with the Illinois General Assembly or Missouri General Assembly, review the ISBE guidance directly, and consult their own counsel before making compliance decisions.

Next step

Bringing this to your school board?

Request the full District Compliance Brief — subprocessor list, data flow diagram, and model disclosure sheet in one document, ready for board review.