Editorial Policy

This is the accountability page for the site. It sets out what I do editorially, what I refuse to do and the standards you are entitled to hold this site to.

Independence

Everything here is written independently of every platform mentioned on the PolyBuzz page. No platform sees its coverage before publication, receives edits in advance or holds a veto, and that includes Character AI, Janitor AI, Chai and Candy AI in the alternatives table. Independence is the working constraint behind how the site runs, not a slogan.

Affiliate revenue in plain language

The site earns affiliate revenue. Some outbound buttons open PolyBuzz or an alternative through an affiliate link, and if you sign up or subscribe after clicking, the destination may pay me a small commission at no extra cost to you.

Affiliate revenue is not allowed to influence any of the following. Each point applies whether or not a commission is attached:

Which platforms appear in a comparison. Selection follows what readers are likely to be asking.
How they are ordered. Order follows fit and price, not payout.
Whether a critical point gets softened. If a platform is worse at something, the page says so.
Whether an alternative gets left out. Every relevant alternative is named, commission or not.

Sources linked without any affiliate arrangement include California's SB 243 text on LegiScan, the New York governor's notice on AI companion safeguards and the FTC's inquiry into AI companion chatbots. They get the same editorial treatment as any affiliate-linked platform.

Corrections and the correction log

Confirmed factual corrections are handled openly. When an error is verified, three things happen: the affected page is edited, its schema dateModified is updated, and a line is added to the log below. Typo fixes and prose polish are not logged, while substantive factual changes are.

Correction log

No entries yet. The first substantive correction will be logged here with the date, the page and the nature of the change.

Sourcing standards

Every material claim on the main page traces to one of five categories: the platform's own materials, an account check, a regulator or public record, a documented user report or a credited third-party write-up. Where a claim rests on inference from those sources, I show the reasoning rather than hiding it behind vague attribution. The weighting between source types is set out on the methodology page.

Use of AI in the editorial process

Language models are used only at the research stage, never as the final drafter of published text. They help compress long sources such as privacy policies and store listings, and every sentence on the main page is written or substantially rewritten by me. The reason is practical: models invent dates, prices and quotes often enough to fail this site's sourcing standard, so nothing reaches readers without a fact-check.

Handling of source citations

Third-party sources are cited directly, by name and with a link where one exists. On the main page that includes Fast.io's breakdown of avatar limits, Bittime's walkthrough of the character builder and the regulator sources on companion chatbots. A source is never paraphrased into saying more than it does, and if a finding comes with a qualification, the qualification travels with it.

Reader rights

You can expect five things from every page: an accurate date, a visible byline linking to the editor's profile, a correction pathway, an affiliate disclosure and the name of the editor responsible. The methodology page covers how those rights work in practice, and this page covers the promises.

Contacting the editor

Editorial matters go straight to me on LinkedIn: linkedin.com/in/melissablake1. That covers corrections, disagreements with a specific claim and press queries, and my response window is seven working days.

Last reviewed: October 6, 2026 · by Melissa Blake

Adult content ahead