Yes — the best comment is to ask Claude to sound like a specific human voice , not just “not AI.” Generic anti-AI prompts usually produce bland output, while a concrete style brief works much better.

What people are saying

A common takeaway from forum-style discussions is:

Don’t tell it what to avoid; tell it what to be.

That means:

  • Give Claude a real audience.
  • Give it a tone.
  • Give it a sample of your own writing.
  • Ask for edits in passes, not one giant rewrite.

Creators discussing this in 2026 also recommend building a repeatable style setup or “voice lock” so Claude keeps the same tone across outputs.

Practical comment you can use

You can comment something like:

Claude output stops sounding AI-generated when you stop asking for “human- like” and start defining a specific voice, audience, and writing rhythm. Give it examples, then revise for clarity, unevenness, and natural phrasing.

Or, shorter:

The trick is specificity. “Don’t sound AI” is too vague; “write like a sharp, slightly casual product manager explaining this to a coworker” works much better.

What works best

  • Use a role, audience, and goal.
  • Add one or two writing examples.
  • Ask for contractions, shorter sentences, and less polish.
  • Request one revision pass focused only on making it sound natural.
  • If possible, draft first yourself and let Claude refine, which many users say preserves a more human voice.

Clean version for a post

If you want a forum-ready comment, use this:

In my experience, Claude sounds most natural when you give it a real voice to imitate, not just a ban on AI clichés. Specificity beats vague instructions every time.

TL;DR

The strongest comment is: “Don’t ask Claude to avoid AI writing — tell it exactly what human voice, audience, and tone to write in.” That approach is echoed across recent guides and discussions about reducing AI-like output.