Priorities Podcast — Summary

Episode Title: Banning 'woke' AI in Idaho
Host: Colin Wood (StateScoop)
Guest: Quinn Annex Reese, Senior Policy Analyst, Center for Democracy and Technology
Date: March 18, 2026

Episode Overview

This episode tackles recently advanced Idaho legislation—House Bill 687—that would prohibit state agencies from obtaining or using AI models deemed to promote principles of diversity, equity, or inclusion (DEI), “transgender ideology,” or “intersectionality.” Host Colin Wood and guest Quinn Annex Reese discuss the language of the bill, its technical and legal implications, and the broader political movement to regulate “woke” AI systems in government technology procurements.

Key Discussion Points & Insights

1. The Bill’s Definitions and Targets

Timestamp: 01:53–05:04

• Slippery Definitions:
  Reese highlights the ambiguity in the bill:

  “A lot of the definitions we're going to be talking about are very slippery ... there isn’t necessarily an agreed upon definition of, in this instance, woke or what constitutes unbiased. And so it can create really uncertain terrain...” (01:53)

• Contradictory Goals:
  The bill ostensibly aims for “truth seeking” and “ideological neutrality”—which sound reasonable—but directly targets DEI, “transgender ideology,” and “intersectionality” for exclusion, contradicting its neutrality claim.

• Precedent for State AI Governance:
  This marks Idaho’s first legislative foray into AI guidelines for agencies, despite existing resources from the state Office of Information Technology Services.

2. Technical and Practical Challenges

Timestamp: 05:04–10:40

• Enforcement Problems:
  Wood questions enforceability:

  “Even if this did pass, I don't see how this would be enforceable technologically.” (05:04)

• Ideological Neutrality is Undefined:
  Reese points out:

  “Defining neutrality as excluding certain viewpoints makes it inherently political and not neutral. And secondly, there's no single definition that exists for ideological neutrality.” (05:51)
  She further explains that any technical fix (presenting both sides, middle-ground-answers, etc.) has fundamental flaws—e.g., “both sides” logic could legitimize fringe views.

• No Benchmarks:

  “There is no consensus in the field about what type of benchmark you should be using to actually measure neutrality in a system.” (05:51)
  This creates barriers for honest AI procurement.

• Comparison to Accessibility:
  Wood notes state struggles in evaluating technology even with clear standards (e.g., web accessibility), intensifying skepticism about applying subjective neutrality requirements to AI (07:53).

• Resource Burden:

  “...you're either going to need your current personnel to spend a lot of time ... or bring on additional technical expertise... Either way, this is going to be a burden for agencies...” (09:13)
  Reese also casts doubt on the bill’s fiscal note, which claims zero additional implementation cost.

3. Political Context & Federal Parallels

Timestamp: 10:40–13:59

• Federal Executive Order:
  The Idaho bill closely mirrors a 2025 Trump Administration federal executive order—“Preventing Woke AI in the federal government”—which mandated ideological neutrality and truth-seeking for federal agency AI procurements.

• Loose Federal Guidance:
  Federal requirements remain vague, with agencies left to interpret implementation, raising concerns about transferability to resource-strapped state agencies.

• Legal, Procurement, and Innovation Risks:
  The potential for lawsuits over “unbiased” requirements is high; lack of clear technical rubrics could chill innovation or deter vendor participation.

4. First Amendment & Legal Risks

Timestamp: 13:59–16:10

• State AG Actions:
  Some state attorneys general have threatened AI vendors over outputs they disapprove of politically—e.g., Montana AG criticizing Google for “hedged” answers, and Missouri AG objecting to chatbots ranking Trump last.

  “Viewpoint based content restrictions by government actors are unconstitutional.” (13:59)

• Government Contracts as a Loophole:
  While agencies can set contract terms, overreach could affect speech or services for non-government users, threatening access to information and free expression.

5. Projected Consequences for Idaho & Other States

Timestamp: 16:10–19:21

• No Implementation Lead Time:
  The bill would go into effect immediately—with no phase-in for agencies—or added funds.

• Scenarios of Noncompliance or Overcompliance:

  “Will ... agencies ... shrug and say, I think what we're using is fine? Will ... general counsel ... say, oh, this makes me afraid about any LLM we're using. Stop all use indefinitely ...?” (16:31)

• Procurement Paralysis:
  Unclear rules could halt LLM adoption or spark vendor lawsuits.

• Alternative Approaches:
  Reese recommends more common-sense regulatory models:

  “There are models ... to make sure ... AI tools ... are accurate ... and not harming our constituents ... It shouldn't be a zero-sum game ...” (18:33)

Notable Quotes & Memorable Moments

• On Contradictory Neutrality:

  “Defining neutrality as excluding certain viewpoints makes it inherently political and not neutral.”
  —Quinn Annex Reese, 05:51

• On Technical Impossibility:

  “Presenting both sides could lend legitimacy to fringe views... like, what would a both sides response be to what happened at the Apollo 11 moon landing?”
  —Quinn Annex Reese, 06:26

• On Resource Strain:

  “I’m a little bit skeptical of that because, really, for this to be implemented, agencies ... have to do these ... compliance reviews ... or bring on additional technical expertise ... Either way, this is going to be a burden...”
  —Quinn Annex Reese, 09:13

• On Legal Risks:

  “Vendors are going to be incentivized to contest those award decisions in court. And if there's not a clear technical rubric... this could be a really big risk...”
  —Quinn Annex Reese, 12:56

• On Broader Impacts:

  “If this ends up passing in Idaho ... it could really make it so that agencies have an extremely hard time doing basic procurements with LLMs...”
  —Quinn Annex Reese, 16:31

Timestamps for Important Segments

• 01:53 — “Slippery” Definitions and Subjectivity
• 03:02 — Summary of Idaho House Bill 687
• 05:04 — Feasibility & Technical Implementation Challenges
• 07:53 — State Tech Procurement: Accessibility Parallels
• 09:13 — Personnel, Cost, and Infrastructure Concerns
• 11:09 — Federal Executive Order Parallels
• 13:59 — State AGs, First Amendment Litigation, & Vendor Impact
• 16:10 — Immediate Implementation & Agency Impact
• 18:33 — Alternative, Pragmatic Models for AI Oversight

Summary Takeaways

• Unclear Definitions: The effort to legislatively ban “woke” AI (DEI, “transgender ideology,” etc.) is rife with ambiguities and contradicts claims of unbiased neutrality.
• Policy Copycats: Idaho’s bill is modeled on a recent (2025) federal executive order, but states face deeper resource, expertise, and legal constraints.
• Procurement, Legal, & Innovation Risks: Without field-wide standards, such laws risk confusion, legal disputes, procurement paralysis, and stunted innovation.
• Better Alternatives Available: Rigorous, technical, and non-ideological safeguards already exist and offer more practical starting points for state oversight of AI.
• Broader Trend: The Idaho case is part of a wider national flashpoint over AI regulation, bias claims, and the politicization of technology in public service.

This episode provides a thorough, critical look at the limitations, risks, and possible unintended consequences of banning “woke” AI in Idaho—offering insights for policymakers, IT practitioners, and anyone interested in the intersection of technology, law, and politics.