AI Policy

How Matters Under Review uses AI — and how we work to protect the communities affected by it.

Purpose & Scope

Matters Under Review (MUR) publishes this policy for two reasons. First, to disclose how we use AI tools in producing and publishing content on this site. Second, to state our position on how AI affects residents, workers, and civic institutions — because transparency is not only about our own practice, it is also about the systems we depend on.

This policy applies to every page, media file, metadata block, and download published on mattersunderreview.org. It is written in plain language. If a claim is unclear, we consider that a defect and will correct it.

MUR uses AI to support human work — never to replace human judgment on matters of fact, editorial selection, or civic accountability.

How MUR Uses AI

Where AI Is Used

  • Metadata generation. Schema.org JSON-LD, AI routing tags (mur-*), and social preview cards are assembled with AI-assisted tooling and then reviewed by a human editor before publication.
  • Audio narration. Some pages include a "Media Overview" and "Podcast Overview" block. These narrations are synthesized with AI voice tools. They are labeled on the page as "Narration synthesized with AI."
  • Internal drafting support. AI tools may assist with outlining, formatting, or restructuring internal drafts before a human rewrites and fact-checks the final text.
  • Search indexing. An internal search index (search-index.json) is generated programmatically to help readers find content. This index is built from published, human-reviewed pages.

Where AI Is Not Used

  • Editorial judgment. What MUR publishes, and why, is decided by a human editor.
  • Fact verification. Every factual claim published on MUR is verified by a human against a primary source or an authoritative public record.
  • Case record findings. Evidence pages — the site's documented case records — are compiled and verified by a human researcher. AI is not used to interpret findings, draw conclusions, or attribute fault.
  • Source selection. Which public records, documents, or officials are cited is a human decision.

Human Review Guarantee

Every page published on MUR is reviewed and approved by a named human editor before it goes live. The editor is identified in the footer as "Written and fact-checked by Murray Wennerlund." If a page was drafted with AI assistance and then revised and approved by a human, that is disclosed at the page level via the ai-generated meta tag and, where relevant, in the footer.

Disclosure Layers

MUR discloses AI involvement at three layers so that both readers and machines can find it:

  1. Page metadata. A <meta name="ai-generated"> tag on every page declares the level of AI involvement for that specific page.
  2. Footer notice. A plain-language AI disclosure appears in the footer of every page.
  3. This policy. The full statement of practice — the page you are reading now.

AI Involvement Levels Used by MUR

Every MUR page declares one of the following levels in its metadata. The definitions below match what the site publishes.

No

Fully human-authored.

Written, edited, and fact-checked by a human. No AI was used in creating the text. This is the default level for MUR pages.

AI-Assisted

Human-authored, AI-edited.

A human wrote the content; AI tools were used for editing, grammar, restructuring, or summarization. No original claims or arguments were generated by AI.

AI-Generated

AI-drafted, human-reviewed.

AI produced the draft from a human prompt; a human editor reviewed, corrected, and approved the final version before publication.

MUR does not publish content at the level of autonomous — content generated and published with no human review. Doing so would contradict this policy.

How MUR Respects AI-Affected Communities

MUR is an educational platform. But we also recognize that AI is changing the conditions under which residents engage with their own government. Our position on AI in civic life is grounded in the same values that shape our educational content: discipline, evidence, transparency, and respect for the people most affected by a decision.

What MUR Supports

  • Disclosure of AI use in public decision-making. When a local government uses AI to triage, score, or recommend decisions affecting residents, that use should be disclosed and reviewable.
  • Human review rights. Residents affected by an AI-driven public decision should have access to a human review process.
  • Privacy of personal data. Public records belong to the public — but personal data should not be scraped or resold without legal authority.
  • Worker protections. AI's impact on public-sector employment should be documented and managed transparently, not silently.
  • Access to understandable language. Tools that translate government documents into plain language serve the public interest. Tools that hide or obscure public records do not.

What MUR Opposes

  • AI-generated disinformation in civic discourse. Synthetic content that impersonates officials, fabricates records, or distorts public meetings undermines the transparency MUR exists to support.
  • Silent deployment of AI in public services. A decision that affects a resident's zoning permit, benefits claim, or police record should not be made by an undisclosed algorithm.
  • Unauthorized scraping of personal data. AI systems may not use personal information published on MUR — or referenced by MUR — to train models without legal authority.
  • AI used to obscure public records. Redaction tools, automated denials, and AI-generated "response summaries" that replace actual records defeat the purpose of public oversight.
  • Claims of AI authority over law. No AI system decides what the law is. Interpretation belongs to courts, legislators, and the public record.

AI and MUR's Own Role

MUR uses AI to amplify civic education, not to replace human judgment. The tools we use are assistants, not authors. Where AI helps produce a page, a human editor is named and accountable. Where AI produces narration, the page says so. Where AI is not used, this policy says that too.

We hold ourselves to the standard we ask of public institutions: use AI where it helps, disclose it where it matters, and keep a human answerable for the outcome.

Accountability & Contact

If you believe an MUR page has misrepresented AI involvement, published an inaccurate AI-generated claim, or used AI in a way that conflicts with this policy, please report it. Reports are read by the editor and addressed in writing.

Policy review schedule. This policy is reviewed at least annually, and whenever the tools MUR uses materially change. Material changes are noted in the footer of affected pages and in the site's news section.

Glossary Term: Human-Authored Content

Definition: Content that was written, edited, and verified entirely by a human, with no AI involvement in generating the text.

In Practice: At MUR, this is the default level for published pages. The author is named in the footer, and the page metadata declares ai-generated: no.

Why It Matters: Human authorship is the foundation of accountability. When a named person stands behind a claim, readers can evaluate the claim against a real, responsible source.

Glossary Term: AI-Assisted Content

Definition: Content authored by a human but refined with AI tools for editing, grammar, restructuring, or summarizing. AI did not generate original claims or arguments.

In Practice: At MUR, AI-assisted work is disclosed at the page level when it materially affects the final text. Editorial judgment and fact verification remain human.

Why It Matters: Most "AI in publishing" debates blur the line between assistance and authorship. Naming the difference makes disclosure honest instead of rhetorical.

Glossary Term: AI-Generated Content

Definition: Content drafted by an AI system from a human prompt, then reviewed and approved by a human editor before publication.

In Practice: At MUR, AI-generated drafts are not published until a named human editor has verified every factual claim and approved the final text.

Why It Matters: A draft is not a source. Review and approval are the acts that convert AI output into a publishable claim.

Glossary Term: Synthesized Narration

Definition: Audio narration produced by text-to-speech or voice-synthesis tools.

In Practice: At MUR, "Media Overview" and "Podcast Overview" blocks use synthesized narration. Every such block is labeled "Narration synthesized with AI."

Why It Matters: Voice is not neutral. Listeners deserve to know whether the voice they hear belongs to a person, and if not, what tool produced it.

Research Resource

Share This Resource

Back to Top