AI transparency
Updated on October 27, 2026
Hypepen uses a writing model to suggest drafts. Here is how, what stays under your control and the limits of these suggestions.
What Hypepen writes
Drafts of posts, threads and articles, hints for replying to mentions (one sentence of advice, never the text of a reply), edits, ideas and remixes of past posts are suggested by a writing model (OpenAI, or xAI depending on the model in use), based on your posts, brand, notebook and instructions.
Until you approve it, a text written by Hypepen carries the “Hypepen draft” badge.
What stays under your control
Nothing is published without your action, except on a slot you set to automatic publishing yourself, which you can turn off at any time.
You review, edit and publish under your editorial responsibility. A suggested text may contain errors: check facts and numbers before publishing.
Estimates
Best times, view and click forecasts, Lab verdicts and recommendations are calculated from your past results, with cautious methods. They are estimates, not promises.
Illustrations
Illustrations are generated by an image model (OpenAI). File metadata is removed before publishing; an invisible watermark may remain in the image.
When you add an illustration, the box “Tell X the image is AI-generated” is ticked. Unticking it is your responsibility.
What we never do
Train a model with your posts, your Content or your X data.
Reply to someone who did not mention you, like, follow or send a direct message on your behalf.