Content Policy

Threadmill generates only text posts and replies for a Threads account the customer personally owns and has authenticated. We do not generate images, audio, or video, and we do not imitate the likeness or voice of any real person.

Generation is grounded solely in the persona the customer configures and the knowledge materials they upload themselves. We do not scrape and redistribute third-party content.

Every draft passes automated screening before it can be published: classification against prohibited categories (hate, harassment, sexual content, false medical or financial advice, illegal activity), per-account blocklists, and duplicate detection against recent posts. Drafts that fail are discarded and regenerated from a different angle.

Links are capped at five per post, and drafts exceeding the 500-character limit are not published.

Approval mode is the default: nothing publishes until the customer reviews and approves the draft in the dashboard, by email, or via Telegram.

Autopilot is an opt-in setting the customer enables per account. Even with autopilot on, the automated screening in section 2 applies identically. Customers can pause an account or disconnect it at any time.

A hard cap of five posts per account per day applies regardless of plan. Posts are spaced at human-like intervals, and we check the account's publishing allowance through the API before publishing to avoid exceeding it.

Similarity checks run within and across accounts so the same wording is not published repeatedly.

Operating accounts owned by others; creating, buying, or selling accounts; manipulating likes, follows, views, or any other engagement signal; spam or bulk promotion; and generating content that uses someone else's copyrighted work or likeness without authorization are all prohibited.

Confirmed violations result in restricted service, with prior notice where practicable.

The account owner is ultimately responsible for what is published. We do not warrant the accuracy of generated output, and customers always retain a means to review before publishing.

Report policy-violating content: [email protected]