Pre-Screening and Post-Screening Policy
Last updated 17 May 2026
Effective date 1 January 2026
We apply review to user content and AI-generated outputs both before they are made available to others on the Service ("pre-screening") and on an ongoing basis after publication ("post-screening"). This policy describes how those processes work.
1. Pre-screening
Public content — including published characters, public profile media, and content submitted to community features — is screened before it becomes available to other users. Pre-screening combines automated classifiers with human review of items that the classifiers flag or that fall into categories requiring stricter review.
2. Generation-time filtering
When you generate content, prompts and outputs are filtered against our prohibited categories in real time. Generations that match prohibited categories are blocked and may be logged for safety analysis.
3. Post-screening
Once published, content remains subject to ongoing review. We may re-evaluate any content at any time — for example, in response to new reports, updates to our classifiers, changes in law, or routine sampling — and remove or restrict content that no longer complies with our policies.
4. Identification of depicted persons
Where content appears to depict an identifiable real person, additional checks may apply, including blocking publication until consent or proof of fictional origin can be established. Suspected depictions of minors are blocked, removed, and escalated.
5. Records and audits
We retain logs of pre- and post-screening decisions, sufficient to support audits, internal quality control, and responses to lawful requests, while minimising personal data in line with our Privacy Policy.
Contact us
If you have questions about this policy, contact AUVORA HOLDINGS LTD at [email protected].