Guides and resources

Claude now marks the text it generates, and OpenAI has said that it will follow. We look at how that changes things for anyone who drafts with a model.

The habits that make model drafts sound generic, and how to get your organization’s voice back.

What we do to a model’s draft, layer by layer.

Fabricated and misdescribed references are the most damaging flaw in AI-assisted drafts, and the easiest to check for once you know where to look.

Why we refuse work aimed at lowering AI detection scores, what the tools actually measure, and the standard you should hold your writing to instead.