Moderation & Reporting

DS Colony · Last updated 1 October 2026

[Beta draft — flagged for legal counsel review. This is the public summary; the internal operator runbook lives with the platform operator.]

1. How it works, honestly

Moderation on DS Colony is human. There is no AI classification, no keyword scanning, and no third-party moderation service — reports go to a queue that a person reads and acts on. The single exception is deterministic: reports in the child-safety category hide the content immediately on receipt, before human review, because that trade is correct. Every moderation action is logged with who, what, when, and why.

We do not proactively monitor everything — we can't and don't claim to. Enforcement is reactive (reports) plus brakes (rate limits). At closed-beta scale with an engaged operator, that is a deliberate and sufficient posture.

2. What you can report, and how

Discussion posts, comments, direct messages, and accounts. Every surface carries a Report control. Reports are private: the person you report is never told who reported them. One open report per person per item — repeats are acknowledged, they don't stack. Filing frivolous or abusive reports is itself a violation.

3. Report categories

• Harassment or bullying — Targeted mistreatment of a person.
• Threats or violence — Threats of harm to anyone.
• Hate speech or abuse — Content attacking people or groups.
• Sexual or explicit content — Not permitted anywhere on the platform.
• Child safety (CSAM) — Highest priority: hidden immediately on report, escalated to I4C, evidence preserved.
• Doxxing / private information — Someone else's private data posted without consent.
• Spam or mass promotion — Unsolicited bulk promotion.
• Scam or fraud — Deceptive schemes.
• Phishing or malicious links — Links intended to steal or harm.
• Impersonation — Pretending to be another person or entity.
• Malware or malicious payload — Distribution of harmful code or files.
• Illegal content — Unlawful under Indian law.
• Copyright infringement — Unauthorized distribution of protected work (see the Copyright Policy).
• Credentials or secrets — Shared logins, API keys, passwords.
• Deceptive content — Misleading or fraudulent claims.
• Off-topic or political — Campaigning or content outside the platform's purpose.
• Something else — A human reviews it either way.

4. The enforcement ladder

After human review, a report resolves as one of:

• Dismiss — no violation found; nothing changes.
• Hide — the content is removed from public view. This is non-destructive and reversible: the row is preserved (evidence, appeal, legal obligations) but never served to users again unless restored.
• Warn — the account receives a clear notification about the reviewed content and the rules; repeat violations escalate.
• Suspend — login is refused with an honest reason, all sessions are revoked immediately, and the account can be reinstated after appeal. Suspension is for serious or repeated violations.

Account actions never happen through an automated decision. Admins cannot be suspended through the queue. Every action also writes an immutable audit row (moderator, action, target, reason, time).

5. Response targets

Reports: acknowledged instantly, reviewed with a 24-hour internal target. Formal grievances (any complaint about the platform or a user): acknowledged within 24 hours, resolved within 15 days by the Grievance Officer — Ankush Jha ([email protected]). Court orders / government directions to remove content: acted on within 36 hours.

6. Appeals

If your content was hidden or your account was suspended and you disagree, email the Grievance Officer with the details. Appeals are re-reviewed by a human; hidden content can be restored, suspensions can be lifted. Appeal outcomes are logged.

7. Transparency notes

Reporter identities are visible only to the moderation queue (to police report-abuse), never to reported users. Reported message content is readable only by conversation members and, for the specific reported message, the queue — that path is role-gated and logged. Moderation records survive account deletion with identities stripped: the audit trail persists, the personal data does not.