Part of the AI Regulation News hub.
A proposed Education Department accreditation rule would require accreditors to check that institutions maintain research-misconduct policies, and its preamble tells them artificial intelligence must not be used by faculty to supplant teaching and instruction
The regulatory text does not mention artificial intelligence once. The preamble does, three times, and it is the preamble that tells accrediting agencies what a compliant institutional policy on faculty AI use looks like. That gap is the whole story for anyone deciding what they actually have to do.
Bottom line: A notice of proposed rulemaking, open for comment. The operative proposal is that an accrediting agency must evaluate whether an institution maintains policies related to research misconduct. The Department's expectations about faculty use of AI appear in the preamble, not in the proposed regulation.
Who this affects: Accrediting agencies recognised by the Secretary, provosts and research-integrity offices at institutions participating in Title IV programmes, general counsel advising on faculty policy, and anyone drafting an institutional AI-use policy for teaching or research.
Issue date: Published 20 August 2026 at 91 FR 53940. Comments must be received on or before 21 September 2026.
What changed: Nothing yet. If finalised, proposed section 602.17(a)(2)(vi) would add research-misconduct policies to what an accreditor must evaluate when assessing institutional compliance with its standards.
Analysis: Read where the AI sentences sit. They are the Department's stated beliefs about what those research-misconduct policies should cover, offered while explaining the proposal. That gives them interpretive weight with accreditors without making AI use a regulatory standard in its own right, and it means a comment arguing about AI is really a comment about the scope of research misconduct.
Primary sources: Notice of proposed rulemaking, official GPO text (91 FR 53940) · Same document, GPO PDF
- Instrument
- Accreditation, Innovation, and Modernization: The Secretary's Recognition of Accrediting Agencies; Institutional Eligibility Under the Higher Education Act of 1965, as Amended; Student Assistance General Provisions
- Authority
- Office of Postsecondary Education, US Department of Education
- Citation
- 91 FR 53940, pages 53939 to 54021, published 20 August 2026
- Regulatory identifiers
- 34 CFR Parts 600, 602 and 668; Docket ID ED-2025-OPE-1042; RIN 1840-AD82
- Jurisdiction
- United States, federal
- Status
- Notice of proposed rulemaking. Open for comment.
- Bindingness
- None yet. A proposed rule. The AI statements sit in the preamble, which does not bind even if the rule is finalised as drafted, though it is the Department's stated interpretation.
- Issue date / next deadline
- Published 20 August 2026. Comments must be received on or before 21 September 2026.
- Implements
- Executive Order 14279, Reforming Accreditation to Strengthen Higher Education
- Editorial Note
- Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
- Primary source
- https://www.govinfo.gov/content/pkg/FR-2026-08-20/html/2026-17001.htm
What the rule actually proposes
The Department proposes to revise the accrediting agency recognition regulations at 34 CFR part 602 to implement Executive Order 14279, Reforming Accreditation to Strengthen Higher Education, along with other Administration priorities, to align the regulations more closely with statute and to reduce regulatory burden.
The provision that carries the AI discussion is proposed section 602.17(a)(2)(vi). It would require that, in evaluating whether an institution complies with an agency's standards, the agency must evaluate whether the institution maintains policies related to research misconduct.
The Department describes what research misconduct covers: it includes, but is not limited to, plagiarism and material misrepresentation of research findings, including through undisclosed selective reporting or other practices that distort the accuracy or reliability of results.
That is the regulatory hook. Everything the Department says about artificial intelligence hangs off it.
The three AI sentences, and where they sit
In explaining the proposal, the Department states that accrediting agencies must ensure institutions have policies in place to prevent faculty members from engaging in improper authorship attribution, citation manipulation, or coordinated practices intended to inflate or misrepresent scholarly impact.
Then the AI passage, in full. The Department encourages the use of artificial intelligence to enhance productivity, streamline processes, and make teaching more efficient. It also believes agencies and institutions must ensure that artificial intelligence is not used by faculty to supplant teaching and instruction. And it does not believe that artificial intelligence should degrade methods in the conduct and presentation of research.
It closes the thought by saying the proposed regulations help to maintain the integrity of research and among the institution's faculty by encouraging responsible uses of beneficial technology.
The framing is deliberately two-sided: encouragement for productivity and efficiency, a limit at supplanting teaching, and a quality floor for research methods. None of those three are drafted as regulatory standards.
Why the preamble placement matters
A preamble explains an agency's reasoning. It does not carry the force of the regulation, and it is not what an accreditor is audited against. The proposed rule's text requires institutions to maintain research-misconduct policies; it does not require those policies to address artificial intelligence.
The practical effect is nonetheless real. When an accrediting agency asks what a compliant research-misconduct policy looks like, the Department has now put on the record that it expects such policies to reach faculty AI use in research and, separately, that AI should not supplant teaching. Agencies tend to build review criteria from exactly this kind of statement.
For an institution the drafting consequence is specific. A research-integrity policy that addresses plagiarism, authorship and data reporting but says nothing about generative tools will satisfy the proposed regulatory text as written, while sitting outside what the preamble describes.
For a commenter the consequence is sharper still. Because the AI expectations are attached to the meaning of research misconduct rather than to a standalone AI standard, a comment about AI is in substance a comment about how broadly section 602.17(a)(2)(vi) should be read. Comments framed as being about AI policy in the abstract are likely to land less well than comments about the scope of that provision.
The academic freedom provision alongside it
The same part of the rulemaking deals with academic freedom, and the drafting history is instructive about how the Department is using non-binding text generally.
The Department had considered defining academic freedom in the regulations, then removed the definition from the proposal. Instead it proposes to require that agencies adopt and implement standards for evaluating whether an institution has an academic freedom policy, and it sets out its own definition as non-binding reference language.
It is explicit that it is not requiring agencies or institutions to apply that definition, while stating that an agency or institution applying it would meet the requirements of the Executive Order and the proposed regulations.
That is the same technique as the AI passage: a substantive expectation expressed outside the regulatory text, with a safe harbour for anyone who adopts it. Institutions reading this rulemaking should treat the non-binding material as the Department's compliance template rather than as commentary.
What to file, and by when
Comments must be received on or before 21 September 2026, submitted through the Federal eRulemaking Portal at regulations.gov under Docket ID ED-2025-OPE-1042. The Department will not accept comments by fax or email, nor comments submitted after the comment period closes, and asks that the Docket ID appear at the top of the comment.
A plain language summary of the rule is available on regulations.gov, as required by the Administrative Procedure Act at 5 U.S.C. 553(b)(4).
One practical note on dates. The Federal Register text sets the deadline at 21 September 2026; the regulations.gov docket record displays 22 September. The published rule is the operative statement, so work to 21 September.
The rule is long, running from page 53939 to 54021, and the AI material occupies a few sentences within it. Anyone commenting on that material should cite the section it attaches to rather than the preamble alone.
Do not report or plan around this as a federal AI rule for universities, because the proposed regulatory text does not mention AI at all. What exists is a proposed requirement that accreditors check institutions maintain research-misconduct policies, plus a preamble in which the Department says it expects those policies to keep AI from supplanting teaching and from degrading research methods, while encouraging AI for productivity. Three consequences follow. Institutions should extend research-integrity policies to generative tools now, since that is what accreditors will be told good looks like. Accrediting agencies should expect to justify their review criteria against preamble language that is not itself a standard. And anyone commenting by 21 September 2026 should argue about the scope of proposed section 602.17(a)(2)(vi), because that is the provision the AI expectations are attached to.
Source File
https://www.govinfo.gov/content/pkg/FR-2026-08-20/html/2026-17001.htm
Open the GPO text and confirm four things: the header showing 34 CFR Parts 600, 602 and 668 with Docket ID ED-2025-OPE-1042 and RIN 1840-AD82, published 20 August 2026; the DATES line setting 21 September 2026; the SUMMARY naming Executive Order 14279; and the discussion of proposed section 602.17(a)(2)(vi), where the sentences about artificial intelligence appear immediately after the description of research misconduct.
We encourage the use of artificial intelligence to enhance productivity, streamline processes, and make teaching more efficient. However, we also believe agencies and institutions must ensure that artificial intelligence is not used by faculty to supplant teaching and instruction. US Department of Education, Accreditation NPRM, 91 FR 53940, 20 August 2026
FAQ
Does this rule regulate AI?
Not in its text. The proposed regulation requires accrediting agencies to evaluate whether an institution maintains policies related to research misconduct. The statements about artificial intelligence appear in the preamble, which explains the Department's reasoning and does not carry the force of the regulation.
What exactly does the Department say about AI?
That it encourages the use of AI to enhance productivity, streamline processes and make teaching more efficient; that agencies and institutions must ensure AI is not used by faculty to supplant teaching and instruction; and that it does not believe AI should degrade methods in the conduct and presentation of research.
Which provision do the AI statements attach to?
Proposed section 602.17(a)(2)(vi), under which an accrediting agency evaluating institutional compliance must evaluate whether the institution maintains policies related to research misconduct.
What counts as research misconduct here?
The Department says it includes, but is not limited to, plagiarism and material misrepresentation of research findings, including through undisclosed selective reporting or other practices that distort the accuracy or reliability of results. It adds that agencies must ensure institutions have policies preventing improper authorship attribution, citation manipulation, or coordinated practices intended to inflate or misrepresent scholarly impact.
When is the comment deadline?
The Federal Register text requires comments on or before 21 September 2026, through regulations.gov under Docket ID ED-2025-OPE-1042. The regulations.gov docket record displays 22 September; the published rule governs.
What is Executive Order 14279?
Reforming Accreditation to Strengthen Higher Education. The NPRM proposes to revise the accrediting agency recognition regulations at 34 CFR part 602 to implement its directives along with other Administration priorities.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.