AI in Schools / State report
New York state policy and NYC AI rules
The August 23, 2026 tracker listed New York without a statewide classroom-AI rule. NYC’s March guidance sets separate local limits.
Published
Research as of · Sources & Notes
No statewide rule in the August snapshot
New York, Pennsylvania, Texas and DC: State snapshot and local rules
In the August snapshot, these four were in the no-statewide-rule category. Yet New York City requires AI tools to pass review before use with student data, Philadelphia names district-approved products, Houston conditions older students' access on written parent consent and task-specific teacher permission, and DCPS prescribes approved-tool and staff-use safeguards. Students and families must identify their own district, rather than borrow a neighboring district's permission. Educators and leaders should make the local tool list and use conditions accessible. Washington and Massachusetts, by contrast, sit in the tracker guidance category, demonstrating why 'state guidance' and 'local rule' need separate fields. [7][8][42][34][14]
NYC requires product review before student-data use
New York City's March guidance makes vendor review the first gate: an AI tool cannot be used with student data before ERMA review, and the ten-step process covers implementation. The city bars student-data model training and reserves high-stakes decisions for people. Parents can request information about AI tools in their child's instruction, but the guidance cited does not establish a blanket parent-consent requirement. Students should ask if the tool and assignment are approved; educators should not enter records in an unreviewed service; leaders need evidence of completed review. [8]
What must NYC review before a tool reaches student data?
NYC reviews products before use with student data
The March guidance requires ERMA review and retains human decisions. It does not establish blanket parent consent.
NYC reviews products before use with student data. The March guidance requires ERMA review and retains human decisions. It does not establish blanket parent consent. ERMA product review. Action: Complete ERMA review before use with student data. Scope or limit: The approved report describes a ten-step implementation process without listing all ten steps. Evidence state: first party guidance. Student-data restrictions. Action: Vendor conditions prohibit training models with student data. Scope or limit: An unreviewed service is not a place to enter student records. Evidence state: first party guidance. Human decisions. Action: Educators and qualified humans retain grading, placement and discipline decisions. Scope or limit: Parents can request information. The cited guidance does not establish blanket parent consent. Evidence state: first party guidance.
ERMA product review
Official guidance
- Action
- Complete ERMA review before use with student data.
- Scope or limit
- The approved report describes a ten-step implementation process without listing all ten steps.
Source notes
Source notes [8]
Student-data restrictions
Official guidance
- Action
- Vendor conditions prohibit training models with student data.
- Scope or limit
- An unreviewed service is not a place to enter student records.
Source notes
Source notes [8]
Human decisions
Official guidance
- Action
- Educators and qualified humans retain grading, placement and discipline decisions.
- Scope or limit
- Parents can request information. The cited guidance does not establish blanket parent consent.
Source notes
Source notes [8]
View data and source notes
| Item | Action | Scope or limit | Status | Sources |
|---|---|---|---|---|
| ERMA product review | Complete ERMA review before use with student data. | The approved report describes a ten-step implementation process without listing all ten steps. | first_party_guidance | [8] |
| Student-data restrictions | Vendor conditions prohibit training models with student data. | An unreviewed service is not a place to enter student records. | first_party_guidance | [8] |
| Human decisions | Educators and qualified humans retain grading, placement and discipline decisions. | Parents can request information. The cited guidance does not establish blanket parent consent. | first_party_guidance | [8] |
Sources & Notes
This report reflects research as of October 6, 2026. Reference numbers link to the cited publications.
The August tracker counts describe its historical snapshot. The reviewed policy database has a separate scope and release date. Neither is a census of classroom AI use.
Limits and source differences
- Classroom AI by State and District | Formative Spaceswww.formativespaces.com
- Guidance on Artificial Intelligencewww.schools.nyc.gov
- Student and Staff Technology and Network Acceptable Use Policydcps.dc.gov
- A I G u i d e b o o khoustonisd.org
- AI Resources – PSTVwww.philasd.org
Share this report
Download a branded preview or share the report link.
Related reporting
California law bars ed-tech use of student data for AI training, extends privacy rules to colleges
AB 1159 prohibits companies from using identifiable student information to train generative AI and broadens California's school privacy laws to cover higher education, closing a loophole that let general platforms escape regulation.
Massachusetts Graduation Council Recommends AI Literacy as Required Milestone
Final advisory report filed June 2026 proposes AI and digital literacy module within career-planning framework; legislative action still needed before any class graduates under new standards.
LAUSD bars student generative AI on district devices for 2026-27, surprising board and parents
The nation's second-largest school district imposed a moratorium on student access to generative AI tools, affecting 378,000 students. The move was disclosed at a committee meeting without a board vote, as the district studies how to regulate the technology.