AI Policy Lab — Policy Repository Workbook

AI Policy Lab · Policy Repository Workbook

Campus Writing and Speaking Program · NC State University · 2026

This workbook is a working document. Write in it. Argue with it. The annotations in Part II model one reading — not the only possible reading. Your analysis in Part III should challenge, extend, or contradict what you find here.

This workbook supports the AI Policy Lab workshop, a faculty professional development experience offered through NC State's Campus Writing and Speaking Program. It has four parts.

Framework

How to use the analytical tools

Corpus

Annotated entries from the corpus

III

Your Analysis

Blank templates for your documents

Draft

Adaptable policy language

Framework: Cole (2026). Three-Level AI Literacy Model. Pedagogies: An International Journal. · go.ncsu.edu/cwsp

Part I

How to Use This Workbook

Every annotated policy entry in Part II applies the same four-step analytical procedure and three-level literacy framework. Apply the same procedure to new documents in Part III.

The Analytical Procedure

Four moves for reading any AI governance document, applied in sequence.

Identify the genre

What type of document is this? Who issued it, and in what institutional capacity? Genre determines what authority the document carries and what it can and cannot require.

Locate the glitch

Where does the document strain, contradict itself, or produce language that hedges rather than governs? The glitch is where the document's work becomes visible.

Read the reveal

What does the glitch expose about what the institution values, fears, or cannot resolve? What structural condition produced this document?

Name the abstraction

What general claim about AI governance, authorship, or power does this document exemplify? What principle could guide better policy at this institution?

The Three-Level Framework

After applying the four-step procedure, characterize the literacy demands the document makes — or fails to make — at three levels.

Conceptual

Recognition and naming: what terms does the policy use, and what do those terms assume? What can students and faculty name, and what remains unnamed?

Structural

Power and infrastructure: whose interests does the governance structure serve? Who authored this document, and what institutional pressures shaped it?

Operative

Contextual judgment: what operative choices does the policy ask faculty and students to make? What decision-making capacity does it cultivate or foreclose?

Reading for the Glitch

Every AI governance document produced under institutional pressure reveals something its authors did not intend. We call this the glitch — the moment where the document's language breaks down, hedges, contradicts itself, or produces an unintended meaning.

The glitch is not a flaw to be corrected. It is an analytical resource. It makes visible what the institution values, what it fears, and what it could not resolve at the time of writing.

Look for: hedging language ("strongly discouraged"), contradictions between sections, silences where governance should be explicit, borrowed frameworks applied beyond their scope, and temporal artifacts — language that was accurate when written but is no longer true.

Part II

Annotated Policy Corpus

This section presents six annotated policy entries selected from the 80-document corpus. Two entries focus on NC State documents analyzed as a depth case; four present comparative institutional examples. Each entry applies the four-step analytical procedure and the three-level literacy framework.

These annotations model one reading — not the only possible reading. Use them as a starting point, not a conclusion. Discussion notes at the end of each entry suggest how to open the annotation up to disagreement.

NC State Depth Case · Entries 1–2 of 4

"any attempt to shortcut that training such as using AI to write an essay is fundamentally at odds with what education is about"

Four-Step Analysis

Genre

Faculty newsletter: an internal advisory document issued by the Office for Faculty Excellence, not the academic policy office. Carries persuasive authority but no enforcement mechanism.

Glitch

The document speaks with the certainty of institutional policy while explicitly deferring to faculty discretion. It issues a moral claim ("fundamentally at odds") without issuing a rule.

Reveal

The institution does not know who is responsible for AI governance. The OFE newsletter fills that vacuum with moral rhetoric rather than policy infrastructure.

Abstraction

When governance infrastructure is absent, moral rhetoric performs its function. The performance reveals the absence.

Three-Level Literacy Read

Conceptual

The document introduces no AI-specific vocabulary. "AI" and "generative AI" are used interchangeably without definition.

Structural

Faculty discretion is encoded as the operative governance mechanism. The OFE, not academic policy, is the authoring body — a structural mismatch.

Operative

The document offers instructors no decision framework. "Faculty discretion" is both the governance solution and the governance problem.

Discussion Notes

Useful entry point for discussing the difference between advisory and policy documents. Ask: what would it take for this document to become policy? Who would have to issue it, and in what form?

Three sample statements ranging from full prohibition to active encouragement; no guidance on choosing between them.

Four-Step Analysis

Genre

Instructor resource guide: a practical toolkit from the teaching technology center. It models syllabus options without endorsing any of them.

Glitch

The three-tier model presents three mutually incompatible theoretical positions as equivalent options. Choosing prohibition vs. encouragement is not a preference; it is an epistemological commitment. The document treats them as stylistic alternatives.

Reveal

The three-tier model reveals that the institution has resolved the question of AI governance by refusing to resolve it. Institutionalized flexibility is institutionalized incoherence.

Abstraction

A resource that presents incompatible positions as equivalent choices does not support decision-making; it defers it to the least-resourced actor (the individual instructor).

Three-Level Literacy Read

Conceptual

The DELTA document provides concrete vocabulary for the three tiers (prohibit, acknowledge, encourage) — the clearest conceptual framework in NC State's Wave 1 corpus.

Structural

DELTA — a teaching technology center — is the authoring body. The document carries pedagogical authority but no governance authority. This structural gap is the glitch.

Operative

The document is explicitly designed to support instructor decision-making but provides no decision criteria. Instructors are given three options and no framework for choosing.

Discussion Notes

Pair with the OFE newsletter. Together they represent NC State's full Wave 1 institutional response: moral rhetoric (OFE) + neutral options (DELTA). Ask: what is missing from both documents that a university-wide policy would provide?

Comparative Institutions · Entries 3–6

Downloadable course-policy icons for three tiers (prohibit / allow with attribution / allow without restriction); recommendation against AI detection tools.

Four-Step Analysis

Genre

Teaching resource with visual design components: a CTI toolkit that includes downloadable assets instructors can embed in syllabi. The icon system converts policy into communication design.

Glitch

Cornell converts governance into a set of downloadable icons — which means enforcement depends entirely on instructors consistently deploying and explaining those icons. The icon is the policy; the icon has no enforcement mechanism.

Reveal

The icon system reveals that governance is increasingly a communication problem, not a rule-making problem. Cornell has resolved the governance question by making it legible — which is not the same as making it enforceable.

Abstraction

When enforcement is absent, governance becomes communication design. Communication design can clarify a position without institutionalizing it.

Three-Level Literacy Read

Conceptual

The icon system is a strong conceptual literacy tool — it gives students a visual vocabulary for understanding different AI use permissions at a glance.

Structural

The CTI is the authoring body — not academic policy, not the faculty senate. The icon system has no institutional authority; its effectiveness depends entirely on voluntary adoption.

Operative

The recommendation against AI detection tools is the most consequential operative move. It shifts the governance burden away from policing and toward communication — which has significant implications for how instructors design assignments.

Discussion Notes

This is the clearest example in the corpus of governance performed as design. Use it to discuss the relationship between legibility and enforceability. Ask: does making a policy visually clear make it more or less likely to function as governance?

Every undergraduate beginning with the class of 2029 required to graduate AI fluent; AI fluency to be embedded across the curriculum rather than siloed in a single course.

Four-Step Analysis

Genre

Policy mandate: an Office of Academic Affairs initiative converting a pedagogical goal into a graduation requirement. Carries enforcement authority that Wave 1 documents typically lacked.

Glitch

The AI Fluency Initiative should be read against OSU's Wave 1 corpus, which includes prohibition-heavy syllabus guidance. The mandate does not acknowledge this reversal, which means it implicitly renders earlier policies incoherent without saying so.

Reveal

The unacknowledged reversal reveals that institutional positions on AI are not stable across waves. Wave 1 documents are now historical artifacts. OSU's failure to acknowledge this makes Wave 2 documents structurally dishonest about what changed and why.

Abstraction

Institutional positions on AI are not stable across waves. The failure to narrate the transition from Wave 1 to Wave 2 is itself a governance problem.

Three-Level Literacy Read

Conceptual

The AI Fluency initiative defines fluency as its own vocabulary: knowing how to use AI, when not to, and how to evaluate AI-generated outputs. This is the strongest conceptual definition in the corpus.

Structural

The mandate is housed in the Office of Academic Affairs — the governance body with authority to require graduation outcomes. This is the most structurally well-placed document in the comparative set.

Operative

The graduation requirement mandate without a defined assessment framework creates a massive operative gap: instructors are required to develop AI fluency but given no shared standard for measuring it.

Discussion Notes

Most productive when read in sequence with OSU Wave 1 entries. Ask: what would an honest narrative of this policy reversal look like? What would an institution need to say to its faculty, students, and the public when it moves from prohibition to mandate?

"Standards explicitly list 'OpenAI's ChatGPT, Microsoft's Bing AI Chatbot, and Google's Bard' as prohibited tools."

Four-Step Analysis

Genre

Student conduct policy update: a School of Law standards revision embedded in the professional conduct framework. The law school context makes named tools a conduct matter, not a pedagogical one.

Glitch

By naming specific tools in a conduct policy, UCLA Law produces a document that is structurally obsolete from the moment it is published. Google's Bard no longer exists; OpenAI's tools have proliferated beyond ChatGPT. The glitch is the temporal artifact problem made explicit.

Reveal

The named-tool policy reveals that Wave 1 governance was calibrated to a specific technological moment — a moment that has already ended. The policy is now a record of what institutions feared in Spring 2023.

Abstraction

Tool-specific governance is structurally unsustainable. Any policy that names tools rather than defining capabilities will require continuous revision — or will go silently obsolete.

Three-Level Literacy Read

Conceptual

The named-tool approach gives students maximum conceptual clarity about what is prohibited — at the cost of a definition that cannot survive tool proliferation.

Structural

The law school issues this document, not the university. Disciplinary specificity means the policy does not generalize — other UCLA schools may operate under different frameworks.

Operative

The operative literacy this document demands from students: know which tools are on the prohibited list. This is compliance, not judgment. Students learn to avoid listed tools, not to reason about AI use.

Discussion Notes

This is the clearest example in the corpus of the temporal artifact problem. Use it to discuss what happens when named-tool policies go unenforced because the tools have changed. Ask: is this document currently being enforced? If so, against which tools?

"promotes the responsible use of artificial intelligence to support and enhance the university's academic, research, and operational goals"

Four-Step Analysis

Genre

Formal institutional regulation: a university-level policy document with legal force, not advisory authority. UF has been building AI infrastructure since 2020 and this is the formal governance document that retrospectively legitimizes it.

Glitch

UF has been building AI infrastructure since 2020 — an NVIDIA supercomputer, an AI university initiative, mandatory AI literacy training. The 2026 formal regulation arrives after the infrastructure, not before it. The policy is retrospective, not prospective.

Reveal

UF reveals that formal governance and institutional practice are not synchronized. The university built its AI infrastructure under informal arrangements; the formal policy legitimizes decisions already made. This is governance as documentation, not governance as decision-making.

Abstraction

Formal policy is often retrospective, not prospective. Across the corpus, institutions tend to formalize positions they have already operationalized — which means formal policy documents describe institutional practice rather than shaping it.

Three-Level Literacy Read

Conceptual

UF's formal regulation uses "responsible use" as its organizing concept without defining what responsible means. The definition is assumed, not constructed — a conceptual gap the document cannot fill.

Structural

The formal policy is housed in university administration — not the AI initiative, not the faculty governance body. This positioning gives it legal authority but separates it from the pedagogical and research communities it purports to govern.

Operative

The operative literacy the document demands is minimal: use AI responsibly. The operative gap is total — no decision framework, no assessment standard, no mechanism for determining what responsible use looks like in specific contexts.

Discussion Notes

Useful for discussing the relationship between governance and practice. UF is exceptional in the corpus for the scale and early date of its AI investment. Ask: what does it mean for a university to have a "responsible use" policy for a system it built, deployed, and requires students to use?

Part III

Policy Analysis Templates

Use these templates during Activity 1, or when analyzing policies you bring from your own institution. Apply the same analytical procedure and three-level framework used in Part II.

Your responses are saved automatically in this browser.

Template 1

Institution

Document type

Date / Wave

Policy stance

Key language

Excerpt or paraphrase — include section reference if available.

Four-Step Analysis

1 Identify the genre

2 Locate the glitch

3 Read the reveal

4 Name the abstraction

Three-Level Literacy Read

Conceptual

Structural

Operative

Discussion notes / what surprised you

Template 2

Institution

Document type

Date / Wave

Policy stance

Key language

Four-Step Analysis

1 Identify the genre

2 Locate the glitch

3 Read the reveal

4 Name the abstraction

Three-Level Literacy Read

Conceptual

Structural

Operative

Discussion notes / what surprised you

Template 3

Institution

Document type

Date / Wave

Policy stance

Key language

Four-Step Analysis

1 Identify the genre

2 Locate the glitch

3 Read the reveal

4 Name the abstraction

Three-Level Literacy Read

Conceptual

Structural

Operative

Discussion notes / what surprised you

Template 4

Institution

Document type

Date / Wave

Policy stance

Key language

Four-Step Analysis

1 Identify the genre

2 Locate the glitch

3 Read the reveal

4 Name the abstraction

Three-Level Literacy Read

Conceptual

Structural

Operative

Discussion notes / what surprised you

Part IV

Adaptable Policy Language

These templates provide starting-point language for AI use policies at three levels. They are built on four principles drawn from the corpus analysis:

1. Scope terms functionally

Define AI by what it does (generates text, produces images, automates tasks), not by product name. Named-tool policies go obsolete in months.

2. Date your policy

Every policy in this corpus became a historical artifact. Acknowledge the conditions of writing and commit to revision.

3. Name the revision trigger

Specify what would prompt a policy update — tool proliferation, new institutional guidance, evidence that the policy is producing the wrong outcomes.

4. Explain the reasoning

Students comply more with policies they understand. Name why the policy takes the position it does.

The templates below are starting points, not models. Replace everything in [brackets]. Edit directly in each template — your changes are saved automatically in this browser. Use the Download PDF button to export your draft.

Course-Level: Syllabus Statement

Course · Syllabus

Bracket guidance: The first bracket requires a functional definition — not a product name. The second and third brackets require you to make a decision about your course, not defer it. The attribution bracket requires you to name what disclosure looks like in practice. The final sentence commits you to revision — build that time into your semester-end review.

Program-Level: Shared Governance Statement

Program · Handbook / Sequence Governance

The [program name] holds that the communication skills central to [disciplinary practice: e.g., engineering writing, clinical documentation, policy analysis] require [describe the specific human judgment the discipline values: e.g., evidence-based reasoning, rhetorical awareness, contextual decision-making] that AI tools do not reliably produce or assess.

AI tools may be used for [program-level permissions: e.g., literature search, data formatting, accessibility support] across all [program name] courses. Their use for [program-level restrictions: e.g., drafting, analysis, argument construction] requires instructor authorization and student disclosure.

Individual instructors retain authority to apply more specific guidance within their courses. This statement does not supersede course-level policy; it provides the shared framework within which course-level decisions are made.

This statement reflects program consensus as of [date]. It will be reviewed at [faculty meeting / curriculum committee] in [semester/year].

Bracket guidance: The program statement does not need to resolve every question — it needs to provide a framework that individual instructors can apply. The key move is naming the discipline-specific literacy the program values. That naming is the governance decision; everything else follows from it.

Department-Level: Institutional Governance Framework

Department · Curriculum Committee / Faculty Governance

The [department name] adopts the following framework for AI use across its courses and programs, effective [date].

Scope: This framework applies to [describe scope: e.g., all undergraduate courses, graduate seminars, faculty research outputs submitted under department auspices].

Definition: For the purposes of this framework, "AI-assisted work" means [functional definition: e.g., any content produced in whole or in part by a generative AI system, where the system's output constitutes substantive portions of the submitted work].

Governance principle: The department affirms that [name the value: e.g., disciplinary judgment, evidentiary reasoning, authorial accountability] is central to [field] and that AI tools [relationship to that value: e.g., may support but do not replace / are in tension with / require explicit instruction in order to be used appropriately].

Authority: Course-level AI policy is the responsibility of the individual instructor, within the framework above. Program-level guidance is the responsibility of [program director / curriculum committee]. This department-level framework is reviewed by [body] on [schedule].

Revision: This framework was adopted in conditions that are already changing. It will be reviewed at [date] and updated as the technological landscape and the department's pedagogical understanding develop.

Bracket guidance: The department framework sets the conditions within which course and program policy operate. Its most important function is naming who has authority at each level and establishing a review mechanism. The corpus shows that departments that lack this framework tend to produce contradictory course-level policies — not because instructors disagree, but because no one has named the shared commitments.

Your edits are saved automatically in this browser.

The policy your students need is one they can reason from, not just comply with. The distinction between a policy that explains its reasoning and one that does not is the difference between governance that builds literacy and governance that performs authority.

Reference

Full Corpus Index

All 80 documents in the corpus. Use this index to locate documents by institution, wave, or policy stance. URLs for each document are available in the full corpus database; contact CWSP for access.

Status: all entries verified or in verification.

Institution	Document	Date	Wave	Stance
American Public University	Official Generative AI Policy (student handbook)	2023	Wave 1	Transparent / ethical use
Carnegie Mellon University	Provost letter to students	Aug 2023	Wave 1	Unauthorized assistance extended to AI
Columbia University	Provost AI policy	2023	Wave 1	Integrity + privacy dual frame
Cornell University	Committee Report: Generative AI for Education	June 2023	Wave 1	Three-tier framework
Cornell University	Task Force Report: Research	Dec 2023	Wave 1	Cultural norms rather than rules
Cornell University	Center for Teaching Innovation guidance	2023–2024	Both	Icon-mediated governance
Duke University	Academic Integrity and AI use statement	2023	Wave 1	Instructor discretion / context-specific
Duke University	AI Governance Framework	2024	Wave 2	Responsible use / transparency required
Emory University	Academic Integrity policy update	2023	Wave 1	Unauthorized assistance extended to AI
Georgetown University	Provost guidance on generative AI	Aug 2023	Wave 1	Faculty discretion / disclosure required
Harvard University	FAS guidance on AI in coursework	2023	Wave 1	Instructor discretion / context-specific
Harvard University	AI task force report	2024	Wave 2	Integration with transparency
Indiana University	Academic misconduct policy update	2023	Wave 1	Prohibition unless authorized
Johns Hopkins University	Academic Integrity policy revision	2023	Wave 1	Unauthorized assistance extended to AI
Massachusetts Institute of Technology	AI policy guidance for instructors	2023	Wave 1	Course-level determination / no university policy
Michigan State University	Generative AI guidance	2023	Wave 1	Instructor discretion
NC State University	OFE Newsletter — Faculty guidance	Aug 2023	Wave 1	Decentralized / faculty discretion
NC State University	DELTA Syllabus Statement Guidance	2023–2025	Both	Three-tier model
NC State University	Student Conduct Code review	2023	Wave 1	Unauthorized assistance extended to AI
NC State University	ai.ncsu.edu resource hub	2024–2025	Wave 2	Pedagogical support / no governance policy
NC State University	Admissions language revision	Oct 2023	Wave 1	Discouragement (removed after public scrutiny)
NC State University	Office of Faculty Excellence FAQ	2024	Wave 2	Guidance without governance authority
NC State University	IT security guidance: third-party AI tools	2024	Wave 2	Data governance / privacy frame
NC State University	Graduate School guidance	2024	Wave 2	Instructor discretion / disclosure recommended
New York University	Academic integrity policy revision	2023	Wave 1	Unauthorized assistance framework
Northwestern University	Provost guidance on AI in coursework	2023	Wave 1	Faculty discretion / context-specific
Ohio State University	Academic misconduct policy revision	2023	Wave 1	Prohibition-heavy / integrity frame
Ohio State University	AI Fluency Initiative mandate	2025	Wave 2	Mandatory integration / graduation requirement
Penn State University	Academic integrity guidance on AI	2023	Wave 1	Instructor discretion / disclosure required
Princeton University	Academic regulations update	2023	Wave 1	Prohibition unless authorized
Purdue University	Provost guidance on generative AI	2023	Wave 1	Three-tier framework
Rutgers University	Academic integrity policy update	2023	Wave 1	Unauthorized assistance extended to AI
Stanford University	Generative AI policy guidance	2023	Wave 1	Instructor discretion / disclosure required
Stanford University	HAI policy brief on educational AI	2024	Wave 2	Research-led / integration recommended
Temple University	Academic integrity policy revision	2023	Wave 1	Unauthorized assistance framework
Tufts University	Provost guidance on AI use	2023	Wave 1	Faculty discretion / transparency required
UC Berkeley	Academic integrity policy update	2023	Wave 1	Prohibition without authorization
UC Berkeley	Center for Teaching and Learning guidance	2024	Wave 2	Contextual integration / disclosure required
UC Davis	Academic misconduct policy revision	2023	Wave 1	Unauthorized assistance extended to AI
UC San Diego	Academic integrity guidance	2023	Wave 1	Instructor discretion
UCLA	UCLA School of Law Student Conduct Standards	2023	Wave 1	Named-tool prohibition
UCLA	College guidance on AI in coursework	2024	Wave 2	Contextual guidance / instructor discretion
University of Chicago	Academic integrity policy update	2023	Wave 1	Unauthorized assistance extended to AI
University of Florida	Formal Regulation: Responsible Use of AI	2026	Wave 2	Responsible use codified
University of Florida	AI University Initiative governance docs	2020–2024	Pre-Wave	Infrastructure-first / governance retrospective
University of Illinois	Provost guidance on academic integrity	2023	Wave 1	Unauthorized assistance framework
University of Maryland	Academic integrity policy revision	2023	Wave 1	Instructor discretion / context-specific
University of Michigan	AI in education guidance	2023	Wave 1	Faculty discretion / transparency required
University of Michigan	AI task force recommendations	2024	Wave 2	Integration with pedagogical rationale
University of Minnesota	Academic integrity policy update	2023	Wave 1	Prohibition unless authorized
University of North Carolina	Provost guidance on generative AI	2023	Wave 1	Faculty discretion / disclosure required
University of Notre Dame	Academic integrity policy revision	2023	Wave 1	Unauthorized assistance extended to AI
University of Oregon	Academic integrity guidance	2023	Wave 1	Instructor discretion
University of Pennsylvania	Provost guidance on AI in coursework	2023	Wave 1	Faculty discretion / context-specific
University of Pittsburgh	Academic integrity policy revision	2023	Wave 1	Unauthorized assistance extended to AI
University of Southern California	Academic integrity policy update	2023	Wave 1	Unauthorized assistance framework
University of Texas at Austin	Provost guidance on AI use	2023	Wave 1	Faculty discretion / disclosure required
University of Texas at Austin	AI governance framework	2024	Wave 2	Responsible use / transparency frame
University of Virginia	Academic integrity guidance on AI	2023	Wave 1	Unauthorized assistance extended to AI
University of Washington	Academic integrity policy revision	2023	Wave 1	Instructor discretion / context-specific
University of Wisconsin	Provost guidance on generative AI	2023	Wave 1	Faculty discretion / three-tier framework
Vanderbilt University	Academic integrity policy update	2023	Wave 1	Unauthorized assistance extended to AI
Virginia Tech	Academic integrity guidance	2023	Wave 1	Instructor discretion
Wake Forest University	Academic integrity policy revision	2023	Wave 1	Unauthorized assistance framework
Washington University in St. Louis	Provost guidance on AI use	2023	Wave 1	Faculty discretion / disclosure required
Yale University	Academic integrity guidance on AI	2023	Wave 1	Instructor discretion / context-specific
Yale University	Poorvu Center AI in teaching resource	2024	Wave 2	Pedagogical integration / transparency frame