Content Moderation Policy

What we will not help produce

• Content that targets, harasses, or threatens identifiable individuals.
• Sexual content involving minors, in any form.
• Detailed instructions whose primary purpose is to facilitate violence, self-harm, or the construction of weapons of mass harm.
• Material designed to defraud, impersonate, or manipulate a specific person or organisation.
• Mass-produced content whose purpose is to deceive search engines, voters, or markets.

How decisions are made

Automated safety filters operate at the model layer. Edge cases are reviewed by the team against the principles in our Responsible AI Policy and Acceptable Use Policy. We prefer narrow, well-explained refusals to broad, opaque ones.

Appeals

If you believe a rewrite was refused in error, you can describe the situation through the contact page. We will not always agree, but we will respond, and we will reconsider where the facts warrant it.