Press kit — for journalists

1 · One-line pitch (paste-ready)

Use this in pieces that need a one-line description of the project. CC BY 4.0 — attribution to "Policy Window" + a link back to policywindow.org/wiki is appreciated.

2 · News-style attribution format

Most news copy uses informal attribution, not bibliographic citations. For in-article attribution:

Example:

For peer-reviewed-style citations, every article surfaces a Cite this widget with BibTeX, RIS, APA, Chicago, Harvard, OSCOLA, and Bluebook formats. See the bottom of any /wiki/[slug] article.

3 · Embed snippet (drop into your CMS)

Every article ships a stripped-down embeddable card at /embed/wiki/[slug]. Iframe snippet:

<iframe
  src="https://policywindow.org/embed/wiki/eu-ai-act"
  width="100%"
  height="600"
  frameborder="0"
  loading="lazy"
  title="Policy Window — EU AI Act"
></iframe>

CSP allows iframe embedding from any origin. The embed page renders the article body without site chrome — suitable for dropping into WordPress, Drupal, custom CMS templates. Replace eu-ai-act with the slug of any article.

4 · Story ideas (curated, dated 2026-05-30)

A rotating list of story hooks we think are under-covered. Updated quarterly. Each comes with a one-line "why this matters" and a starting point in the catalog.

• 64% of the AI-governance coverage matrix is regulator silence

  Of 624 instrument×topic cells tracked, 64% are silent. The story is which topics no regulator has addressed in writing yet — and which jurisdictions are about to move first.

  Start here: /wiki/meta · /wiki/frontier-regulation

• Three thresholding regimes for GPAI, three different definitions

  EU AIA uses 10²⁵ FLOP compute; the (rescinded) US EO 14110 used 10²⁶; China's GenAI Measures use behavioural triggers (public-facing service). Frontier-lab voluntary codes layer on top with capability-evaluation gates. The story is convergence vs divergence at the compute boundary.

  Start here: /wiki/foundation-models

• EU AI Act timeline: prohibited practices live, GPAI live, high-risk system rules land Aug 2026

  The staged effective dates create three distinct compliance regimes operating concurrently. Every deployer of an Annex III system needs to be ready before the Aug 2026 cliff.

  Start here: /wiki/eu-ai-act

• Frontier-lab voluntary commitments vs regulator-mandated obligations

  Anthropic RSP, OpenAI Preparedness, DeepMind Frontier Safety Framework all gate capability deployment on internal evaluations. EU AIA Art. 55 requires similar — but with different thresholds + reporting. The story is whether voluntary leads or follows regulation.

  Start here: /wiki/debates · /wiki/anthropic-rsp

• Comparative jurisdictional memos: ~10,400 possible EU-vs-X-on-topic-Y pages

  Every two-jurisdiction comparison on every topic is auto-rendered as a 1-2 page memo with PDF download. Useful for "what does country X say vs country Y on biometric ID" pieces.

  Start here: /wiki/compare

5 · Embargo policy

Today: no formal embargo system. Catalog updates flip on deploy and are immediately visible at /wiki/changelog (RSS feed at /wiki/changelog/feed). A journalist wanting early notice on a forthcoming article can email press@policywindow.org; we'll coordinate launch timing where it serves the public interest.

A staged-release infrastructure (registered-journalist preview + coordinated launch) is on the Q4 2026 roadmap (see /wiki/roadmap).

6 · Background on us

• What it is: /wiki/methodology — how the catalog is produced (no LLM-generated article prose, every claim primary-sourced).
• Goals + roadmap: /wiki/goals and /wiki/roadmap — what we're trying to do, when, and how we'll measure it.
• Funding: /wiki/funding — who pays for it (currently: founder personal cost; zero paying customers as of 2026-05-29).
• Operating commitments: /wiki/charter — what we will and will not do.
• AI use: /wiki/ai-disclosure — where AI is used, where it is not.
• Editorial board: /wiki/editorial-board — currently in formation, 1 of 6 slots filled (disclosed honestly).