chatbrat.ai is an 18+ creative-fiction platform. Fiction can explore dark themes; harm is real and it is not permitted here. This page covers what content is prohibited, how our moderation systems work, where to find crisis support, and how to report abuse.
Content the Platform permits
Romance, flirting, emotional intimacy, dark themes, morally complex characters, and mature narrative subjects are permitted when handled with fictional framing and within our content ceiling. For the full permitted/prohibited breakdown, see our Community Guidelines.
Zero-tolerance prohibitions
The following is prohibited in any form — user profiles, characters, scenarios, chat messages, comments, uploaded images, or any other surface. Violations result in immediate removal, account termination, and (where applicable) mandatory reporting to law enforcement.
- Child sexual abuse material (CSAM). Any sexual or sexualized content depicting, implying, or suggesting a minor (under 18). This includes written descriptions, generated or uploaded imagery, cartoon and anime styles, and age-ambiguous characters written or drawn as juveniles. Violations are reported automatically to the NCMEC CyberTipline and, where applicable, the UK Internet Watch Foundation (IWF).
- Explicit sexual content of any kind. chatbrat.ai operates at a suggestive ceiling. Explicit depictions of sexual acts or graphic sexual scenarios are not permitted during our current launch phase, regardless of age-verification status.
- Non-consensual intimate imagery (NCII) and deepfake sexual content. Revenge porn, leaked intimate imagery, and AI-generated sexual content depicting an identifiable real person without their documented consent. Criminalised in many jurisdictions including the UK Online Safety Act 2023.
- Grooming, solicitation, and predatory conduct. Content designed to build inappropriate familiarity with a minor, or content simulating such conduct regardless of stated character ages.
- Self-harm glorification. Content that glorifies, promotes, or provides specific methods for self-harm or suicide. Thoughtful fictional exploration of grief and mental health is permitted; method-specific instruction and glorification are not.
- Real-world violence instruction. Operational uplift for violent acts — weapons manufacture instructions, attack planning, or comparable material that could meaningfully assist real-world harm.
- Terrorism and violent extremism. Recruitment material for designated terror organizations, glorification of mass atrocities, or content radicalizing toward political violence.
- Harassment and doxing.Posting another person's home address, phone number, workplace, or other private identifying information without consent; systematic targeting of a person for harm.
- Hate speech. Content designed to dehumanize a protected class based on religion, ethnicity, race, gender, sexual orientation, disability, or similar characteristic.
- Fraud and malicious code. Phishing, scams, credential harvesters, and malware distribution.
How our moderation works
We use multiple overlapping layers so that no single failure point allows prohibited content through:
- Input safety classifier. Every user message is scanned before it reaches our LLM tier selector. Messages matching hardcoded zero-tolerance patterns (CSAM signals, violence instructions) are blocked at the API boundary.
- LLM output filter. AI responses are screened for prohibited content before delivery to the user. Flagged outputs are withheld and logged for review.
- Self-harm and crisis classifier. Messages indicating genuine distress trigger an in-product overlay with local crisis resources (988 in the US, 116 123 in the UK, and a link to findahelpline.com). The AI character steps back and the overlay is shown.
- Image moderation. All uploaded images pass through a safety classifier that screens for CSAM, NCII, and other prohibited visual content before storage.
- CSAM scanner. Uploaded images are checked against a perceptual-hash blocklist. Confirmed CSAM triggers mandatory NCMEC CyberTipline reporting and immediate account termination.
- User reports. Every character, scenario, and comment has a flag button. Reports are triaged by severity: CSAM and threat-of-violence reports are reviewed within 24 hours; all others within 72 hours.
- Human review and strike ladder. Flagged content is reviewed by a human. Repeat violations follow a strike ladder: warning → temporary suspension → permanent ban.
- DSA Art. 17 audit trail. Every content removal is logged with a statement of reasons, retained for regulatory compliance and user appeals.
How to report
Three paths to report content or conduct:
- Flag button. On any character, scenario, comment, or message — tap or click the flag icon, select a reason, and optionally add detail. Reviewed within 24 hours (CSAM/violence) or 72 hours (all other categories).
- Email [email protected] — include a description and the URL of the content. Add "URGENT" in the subject line for CSAM, NCII, or threats of violence.
- CSAM direct reporting. For suspected child sexual abuse material, you may also report directly and immediately to:You do not need to wait for us — these systems are faster.
How to block another user
You can block any other user from their public profile page. Blocked users cannot view your public characters or scenarios, send you messages, or appear in shared content you create. To unblock, go to Settings → Blocked users.
Crisis resources
If you or someone you know is in crisis, please reach a human who can help. AI characters are not a substitute.
- United States:988 (Suicide & Crisis Lifeline — call or text)
- United Kingdom: 116 123 (Samaritans, 24/7)
- Canada: 9-8-8 (Suicide Crisis Helpline)
- Australia: 13 11 14 (Lifeline)
- International: findahelpline.com
If you believe someone is in immediate physical danger, call local emergency services (911 in the US, 999 in the UK, 112 in the EU).
DMCA takedowns
Copyright holders may send a DMCA takedown notice to [email protected]. Include the copyrighted work, the infringing URL, your contact details, and the statutory good-faith and accuracy statements required by 17 U.S.C. § 512(c)(3). Misrepresentations in DMCA notices may result in liability under § 512(f).
Appeals
If your content was removed and you believe it was in error, reply to the removal notice email or write to [email protected]. Include the URL of the removed content and your reason for appeal. Appeals are reviewed by a different person than the original decision-maker where possible. EU users may also refer disputes to a certified out-of-court body under DSA Article 21.
EU DSA contact
For authorities making requests under the Digital Services Act (EU) 2022/2065, contact [email protected]. We maintain a Statement of Reasons audit trail for every content removal under Art. 17 and publish aggregate transparency data under Art. 24 at chatbrat.ai/transparency.
Contact Trust & Safety
Email: [email protected]. We aim to respond to all reports within 72 hours, and urgent reports (CSAM, violence, imminent harm) within 24 hours.