Tailwind
AI Policy Monitor

How higher ed is governing AI.

A living dataset of published AI policies across higher education in 27 countries, classified by stance, maturity, and audience. We monitor what institutions put in writing, so leaders can see how the sector is actually responding. Updated Jun 1, 2026.

6,602 Institutions monitored
13.1% Have a verified AI policy
1,516 Policies classified
Conditional Most common stance

When campuses first put AI policy on paper

First-adoption year for 458 dated policies.

What posture do policies take?

845 classified policies, by stance.

Conditional: 507 (60%) Unclear: 150 (18%) Encouraged: 85 (10%) Prohibitive: 59 (7%) Permissive: 44 (5%) 60% Conditional
  • Conditional 60%
  • Unclear 18%
  • Encouraged 10%
  • Prohibitive 7%
  • Permissive 5%

Who has a policy, by institution type

Share of each type with a verified AI policy.

Public University 23.1%
University (Canada) 20.9%
Community College 16.5%
Private Nonprofit 15.2%
College (Canada) 13.3%
Tribal College 11.4%
Private For-Profit 5.8%

Do policies require a human in the loop?

Share of classified policies that explicitly require human oversight.

Public University 27.5%
College (Canada) 19.5%
University (Canada) 17.1%
Community College 14.9%
Private Nonprofit 14.8%
Private For-Profit 5.3%

Which AI tools do policies actually name?

Number of institutions that name each tool.

ChatGPT / OpenAI 1,062
Gemini / Google 650
Copilot 647
Claude / Anthropic 369
Midjourney 201
Perplexity 189
Stable Diffusion 101
Llama / Meta 49

How far along is the sector?

All 6,602 monitored institutions, by policy maturity.

  • Comprehensive 242
  • Developing 1,532
  • Minimal 1,564
  • No policy yet 3,264

What the stances mean: Conditional — allowed with guardrails · Unclear — no clear position · Encouraged — actively promoted · Prohibitive — restricted or banned · Permissive — broadly allowed

Coverage changelog

29 active monitoring days, 16,183 coverage updates logged. Most recent first.

  • Jun 2 871 updates · 11 regions Pennsylvania, Texas, New York
  • Jun 1 8,952 updates · 94 regions California, Florida, New York
  • May 24 713 updates · 33 regions North Carolina, Washington, Wisconsin
  • May 23 748 updates · 29 regions England, Germany, Netherlands
  • May 20 50 updates · 2 regions Florida, California
  • May 19 4 updates · 1 region California
  • May 12 8 updates · 1 region Wales
  • May 11 41 updates · 3 regions England, Wales, Washington

Method: institutions are drawn from public directories, AI policy pages discovered via search, verified by crawl, and archived in monthly snapshots. Snapshots are classified by an LLM on a locked rubric for stance, framing, and audience. Percentages use classified policies as the denominator unless noted.