Editorial and AI Use Policy

Research and sourcing

Sources support claims

Safety claims should rely on primary or authoritative public sources wherever possible. The publication process compares advice with the actual user decision, removes claims that exceed the source, and avoids statistics that lack clear context. Citing a source does not replace original analysis, a practical method, or a clear explanation of limits.

Who, how, and why

Accurate organizational authorship

Core pages are published under the Crime Prevent Coalition organization byline. The site does not create fictional writers, credentials, first-hand tests, interviews, or field experience. The work exists to help readers make a prevention decision, not to manufacture pages for search queries.

Automation is not evidence

AI tools may support drafting, structure, code, quality checks, and original illustrations. They are not treated as evidence or independent expert review. Factual claims still require source review, and generated visuals are explanatory editorial art rather than documentary proof.

Imported articles

The Docs Library integration verifies the exact domain identity before fetching. It strips duplicate H1 headings, sanitizes HTML, deduplicates slugs and source IDs, downloads cover images with authorization, and schedules entries at least three days apart. New entries default to pending unless an operator deliberately enables automatic approval.

Automatic approval requires topical fit, a real target in the first-party guide or tool library, substantive structure, practical action items, at least two external source hosts, at least one recognized authoritative source, safe image metadata, no promotional tracking parameters, and low similarity to both the core library and the recent briefing archive. Failure of any check keeps the entry pending.

Indexing threshold

Publication is not a promise of search indexing. The workflow is designed to hold weak, duplicate, unsupported, or off-topic contributions for revision. The sitemap includes only approved posts whose publication time has arrived.

Date integrity

Publication and modification dates reflect the publishing record. A draft timestamp cannot predate the public article in sitemap or structured-data signals, and a date is not changed merely to make unchanged content appear fresh.

Scaled content and site reputation

The publication does not accept unrelated white-label sections, paid ranking pages, distributed keyword articles, or location and synonym variants. Imported work must serve the same crime-prevention audience, pass the site's own gate, connect to a first-party resource, and stand on its own value. A feed, AI system, third party, or available keyword is never sufficient reason to publish.

Commercial and reputation safeguards

We do not rent sections to unrelated third parties, sell followed links, repurpose expired-domain reputation, or create location and keyword variants at scale. External links inside imported copy are sanitized and marked nofollow unless editorial review changes them.

Visual policy

The logo, hub illustrations, and social images were created with the Codex/ChatGPT image generator for this project and optimized to WebP. They avoid identifiable real people, police branding, weapons, and fear-based scenes.

How readers can check a page

Use the same four checks we use

• Confirm that the title and first heading describe the same decision.
• Open the named sources and compare them with the claim they support.
• Read the limits before applying a checklist to a different place or legal setting.
• Use the project and correction information to understand who publishes the page and how problems are handled.

Readers can compare current coverage in the complete crime prevention library. New contributed work appears in the crime prevention briefing archive only after it passes the publication gate. The privacy and data-use notice explains the separate browser-based worksheet workflow.