Divine

Community Information

How Divine enforces its rules, and what we will publish about it.

Effective date: May 16, 2026

This page documents how Divine enforces its rules: the categories of harm we act on, the actions available to our trust and safety team, and the statistics we plan to publish on a regular cadence as the platform grows.

Divine is a young product. Until we have meaningful enforcement volume, the headline statistics below are reported as “to be published” rather than fabricated. We will update this page on a regular cadence (initially every six months, then annually) and clearly indicate the reporting period for each metric.

1. Our approach to safety

1.1 Adults only, verified

Every Divine account confirms its user is at least 18 years old and undergoes a real-human verification step. We pair this with device-level signals and review of reports to keep underage accounts off the platform. See our Child Safety policy.

1.2 Layered moderation

We rely on three layers, in order of priority:

  • Pre-publication filters. All photo uploads (profile photos, additional photos, Story images, post images, chat images) are scanned for harmful content before they are visible to other users.
  • In-product reports. One-tap reporting is available on every user, message, post, Story, Space, and Clan. Reports related to child safety or imminent harm jump the queue.
  • Proactive review. Behavioral and content signals surface accounts for human review before any user reports them.

1.3 Real-time voice considerations

Voice match and Spaces audio are not recorded by us. That choice protects user privacy and aligns with two-party consent norms. The trade-off is that we can’t go back and listen to a call after the fact. To compensate, the App provides post-call reporting with the participant ID and call timestamp attached, and we follow up with both parties when needed.

2. Categories of action

Once we begin publishing volume metrics, they will be broken out into the following enforcement categories. The list is roughly ordered by severity:

  • Child Sexual Exploitation (CSAM, grooming, predatory contact)
  • Self-Harm & Suicide content or solicitation
  • Violent threats & incitement
  • Sexual exploitation of adults (non-consensual intimate content, “sextortion”)
  • Sexual or explicit content (per Community Guidelines §2)
  • Harassment, bullying, stalking
  • Hate speech (slurs, dehumanizing language, group-based incitement)
  • Doxxing, intimate-image abuse, privacy violations
  • Fraud & financial exploitation (romance scams, advance-fee, crypto, pig-butchering)
  • Impersonation & deceptive identity (catfishing, fake profile, synthetic-voice deepfake)
  • Spam, automation, platform abuse
  • Underage account / age misrepresentation
  • Recording / unauthorized capture (voice match, Spaces, message)
  • Off-platform solicitation (pushing to other apps for prohibited conduct)

3. What enforcement actions we take

Depending on severity, intent, history, and context, an enforcement action can include:

  • Educational warning.
  • Content removal (single post / message / image / Story).
  • Feature suspension (posting, messaging, calling, Stories) for a defined period.
  • Profile correction (e.g., adjusted age or required re-verification).
  • Temporary account suspension.
  • Permanent account termination.
  • Device-level block to prevent repeat account creation.
  • Report to NCMEC under 18 U.S.C. § 2258A (for child-safety content).
  • Report to local or federal law enforcement (for credible threats and emergencies).

4. Stats we plan to publish

For each reporting period (initially six months, transitioning to annual), we plan to publish the following metrics, by enforcement category:

  • Number of reports received.
  • Number of accounts actioned (suspended or terminated).
  • Number of pieces of content removed.
  • Median and 95th-percentile response time, by severity tier.
  • Number of NCMEC CyberTipline reports submitted.
  • Number of valid legal-process requests received and complied with, broken out by agency type (federal, state, foreign, civil).
  • Number of preservation requests received under 18 U.S.C. § 2703(f).
  • Number of emergency disclosure requests received under 18 U.S.C. § 2702(b)(8) and the share we acted on.
  • Number of accounts proactively detected and terminated (no user report) for underage suspicion.
  • Number of accounts terminated for evading a prior ban.
  • Voice-classifier accuracy. False-accept and false-reject rates of any voice classifier in production, plus precision and recall on labeled samples. We do not run a voice classifier in production today; if and when we ship one, accuracy stats will appear here.
  • Age assurance. Name of the verification vendor (currently Amazon Rekognition); aggregate pass rate; estimated false-acceptance rate (share of users who passed verification but were later determined under 18 through other signals).
  • 30-day active users. A single denominator for the reporting period so that action counts can be normalized.
  • Appeals reversal rate. The share of appealed enforcement actions that we reverse in whole or in part.
  • Geographic breakdown. Action counts and legal-process counts split United States / EEA / United Kingdom / Rest of World.
  • Reviewer team. At GA we will publish team size, median years of trust-and-safety experience, training hours per reviewer per year, languages supported by human review, and the share of actioned decisions reviewed by a human vs. acted on by automation.

5. Statutory disclosure commitments

Several regimes require specific public disclosures from platforms of our scope as we grow. We will reflect each, where applicable, in future editions of this page:

  • EU Digital Services Act (Articles 15 and 24): notice-and-action numbers, median decision time per Member State, trusted-flagger counts, and out-of-court dispute settlement data.
  • Texas SB 12 (HB 18) protections for known minors: relevant disclosures will appear in the annual report.
  • UK Online Safety Act: information about the categories of illegal content we have detected and removed, and our risk-assessment cadence.
  • California AADC (AB 2273): documentation of our age-appropriate-design assessments where applicable.

6. Our service-level targets

These are the response targets we hold our trust & safety team to, today, with the volume of reports we currently receive. We will report against them publicly in future editions of this page.

  • Child-safety reports: initial review within 1 hour, action within 24 hours, NCMEC report when applicable on the same business day.
  • Imminent physical harm: initial review within 1 hour, escalation to law enforcement when warranted on the same day.
  • Sexual content or non-consensual intimate imagery: initial review within 4 hours; content removed on confirmation.
  • Hate, harassment, threats: initial review within 24 hours.
  • Spam, fraud, platform abuse: initial review within 48 hours.
  • Account-recovery and feature-suspension appeals: initial review within 72 hours.

These are targets, not guarantees, and they tighten over time as we build out the safety team and tooling.

7. Working with NCMEC and law enforcement

Tiger LLC is an electronic service provider for purposes of 18 U.S.C. § 2258A. We report apparent child sexual abuse material (CSAM) to the National Center for Missing & Exploited Children (NCMEC) and cooperate with downstream investigations. For details about how law-enforcement agencies can serve process on us, see our Law Enforcement Guide.

8. Working with civil-society partners

We participate where useful with cross-industry initiatives that reduce online harm, including:

  • NCMEC — CyberTipline reporting and Take It Down (intimate-image takedown for under-18 imagery).
  • Tech Coalition — industry collaboration against online child sexual exploitation.
  • StopNCII.org — hash-based detection of non-consensual intimate imagery for adults.
  • The Internet Watch Foundation (IWF) — CSAM URL list (for U.K. compliance).
  • RAINN — education and crisis-line referrals.
  • Connect Safely and Common Sense Media — user education resources.

9. How we prioritize and decide

  • Reports involving credible threats are reviewed before any other category, regardless of report-queue volume.
  • Permanent account terminations for severe abuse are reviewed by a human before they are issued.
  • Active-threat reports tied to a live voice match or Space surface to a faster queue than asynchronous text-message reports.
  • Reversed appeals are audited; recurring reversal patterns drive updates to training, tooling, and policy.
  • Engagement-optimization features that obscure blocks, reports, or safety controls are out of scope for product development.

10. What you can do

  • Read and follow our Community Guidelines.
  • Report bad behavior in-App so it reaches us with the right context.
  • Block first, report after — protecting yourself is the priority.
  • If you’re ever unsure whether to report, err on the side of yes.
  • Send us feedback. Real user reports of bad patterns are how policies get written.

11. Annual reporting cadence

Once Divine is in full general availability, we will publish a formal Annual Community Report at this URL each calendar year. Interim updates will appear here as the data becomes meaningful.

12. Contact

Tiger LLC — Trust & Safety
California, United States
Email: support@divineapp.io

Specific subject lines that go straight to the right reviewer: “Child Safety,” “Safety,” “Appeal,” “Press — Transparency,” “Legal.”