Editorial standards

AI use policy

How I use AI-assisted tools, and the human review that sits above them.

Last updated: September 2026.

How I use AI tools

Here's what I use AI for and what I don't, so you can judge. I may use AI-assisted tools to support drafting, organization, summarization, grammar review, and editorial workflows. AI tools don't independently publish content. Medical content goes through human review and editorial oversight before publication.

What I use AI for

I find these tools genuinely useful for the mechanical parts of writing, so I'll be specific about where they help. I use AI-assisted tools for:

Drafting. Producing a first pass that a human then rewrites, checks, and takes responsibility for.
Organization. Restructuring an outline or reordering sections so a page reads more clearly.
Summarization. Condensing longer material into a tighter form that a human verifies against the source.
Grammar and style. Catching typos, tightening sentences, and smoothing wording.

What I do not use AI for

There are lines I don't cross, because this is mental health-adjacent material and getting it wrong matters. I do not use AI to:

Generate medical claims. Statements about how the mind works or what an approach can do are written and stood behind by a person.
Produce diagnostic content. Nothing on shrinQ diagnoses, and AI is not used to assess a reader or imply one.
Create citations or sources. I don't let a tool invent references. Sources are chosen and checked by a human, in line with my evidence methodology.

What AI does not do

AI does not replace human review, medical judgment, diagnosis, or treatment decisions. It is not used to give you medical advice, to assess your situation, or to make any individual clinical determination. shrinQ is education, not therapy, diagnosis, treatment, crisis care, or medical advice.

Human review of every AI-touched page

Here's the rule that ties it together: if AI touched a page, a person reviewed it. I don't let AI make medical claims or invent citations. Every claim and every citation on shrinQ has been seen by a human, and that human is usually me. There's no path where an AI-assisted draft reaches you without a human reading it, correcting it, and approving it. The person who signs off is responsible for what appears on the page, the same as if they'd written every word themselves.

Human review and editorial oversight

All medical and educational content is reviewed by a qualified person before it is published, under the oversight described in my medical review board and evidence methodology. Where AI tools assist with drafting or editing, the output is checked, corrected, and approved by a human who is responsible for what appears on the page.

Interactive tools and quizzes

The self-checks, worksheets, and other interactive exercises on shrinQ are rule-based educational tools. They do not diagnose, and they do not generate individualized medical conclusions. Any score or summary they show is a general educational reflection of your own answers, not an assessment by a clinician or by an AI system.

How AI systems can use shrinQ

I want this material to reach people wherever they're asking, and more of them are asking an AI assistant than a search box. So the public site is open. Search engines, AI assistants, and AI developers can crawl, cite, and train on anything you can read on shrinq.com without buying the program.

I ask for two things in return. Credit the source as "shrinQ, medically reviewed by Shariq Refai, MD, MBA" with a link to the page. And don't present it as medical advice or a diagnosis, because it isn't one. The paid program stays under its personal license. The full wording is in the terms, and the machine-readable version is at /ai.txt.

None of this changes how your own information is handled. Entries you type into the interactive tools stay out of training, as described below.

Your information and AI

I do not use the personal entries you make in the interactive tools to train AI models. Where third-party services I rely on use automated systems, that's described in my privacy policy.

Updates to this policy

I revisit this policy when the tools change, not on a calendar. As my tools and processes change, I'll update this page and note the date of the most recent change above. If you have a question about how a particular piece of content was produced or reviewed, please see my corrections policy for how to reach me.