Safety

Safety isn't a feature.
It's the architecture.

Every layer of Under Our Wing is designed to protect users — from the first question in screening to the last message in chat. Here's exactly how it works.

Four Layers

A complete safety stack

Most dating apps add safety as a report button. We built safety into every transition point — before a match is made, before photos are shown, before a message can be sent.

🔍
Layer 1 — Screening

AI screening for red flags and violent signals

The screening conversation isn't only measuring compatibility. It's passively monitoring responses for language patterns associated with aggression, coercive control, entitlement, and criminal behavior. Profiles that trigger these signals are flagged before any match is ever surfaced.

🔒
Layer 2 — Mutual Unlock

No contact without mutual consent

Photos are locked. Messaging is locked. Neither opens until both parties actively choose to unlock the match. No unsolicited messages are possible by design — not suppressed by policy, impossible by architecture.

💬
Layer 3 — Chat Monitoring

AI monitors conversations for safety signals

Once a conversation opens, AI safety monitoring runs passively in the background. Messages that include threatening language, harassment patterns, or coercive signals are flagged automatically. Users receive prompt notification and in-platform tools to act.

🛡️
Layer 4 — Reporting & Response

Structured reporting, blocking, and companion support

Every profile and conversation includes structured reporting tools. Blocks are immediate and silent. The in-app Wing companion is available to help users navigate uncomfortable situations or decide on next steps — without judgment.

AI Screening

What the AI looks for before a match is made

The screening process runs two simultaneous analyses: compatibility scoring and safety monitoring. Safety monitoring is passive — users don't see it, but it's always running.

⚠ Yellow Signal

Entitlement Language

Patterns that suggest a belief that others owe them attention, responses, or access. Predictive indicator of pressure and boundary violations.

⚠ Yellow Signal

Coercive Control Patterns

Language suggesting a history of controlling behavior in past relationships — monitoring, isolation, emotional manipulation framed as care.

⚠ Yellow Signal

Hostility & Blame Attribution

Consistent attribution of past relationship failures entirely to others, combined with hostile language toward former partners.

⛔ Red Signal

Violence & Aggression Markers

Direct or indirect references to physical violence, threats, or menacing language. Profiles triggering this signal are flagged immediately for review.

⛔ Red Signal

Criminal Behavior Disclosure

Statements that suggest or imply criminal history related to violence, harassment, or abuse — whether framed as boast, confession, or offhand comment.

⛔ Red Signal

Predatory Pattern Signals

Language patterns consistent with predatory targeting — escalating intimacy rapidly, testing limits, or expressing interest in vulnerability as an opportunity.

Mutual Unlock

Both people choose. Or nothing opens.

👤

You Unlock

+
👤

They Unlock

On most dating platforms, sending a message requires only your own decision. Under Our Wing requires two decisions. Both parties must actively unlock a match before photos become visible and conversation opens.

If you unlock and they haven't, you see a pending state. No notification is sent to them — no pressure, no awareness that you acted first.

If they unlock and you haven't, same thing. Silence until you're ready.

When both unlock, the connection opens simultaneously. Equal ground. This single constraint eliminates the power asymmetry that makes most dating app inboxes unsafe for many users.

Chat Safety

What happens when something feels wrong

The AI monitoring system runs passively on all conversations. If it detects a safety signal, a sequence of escalating responses activates — starting with user notification and ending with account action if the behavior continues.

🔔

Automatic Detection

The AI scans for threatening language, harassment patterns, sexual coercion, and escalating aggression in real time. Detection is automatic — you don't need to be watching for it.

Immediate User Notification

When a safety signal is detected, the affected user receives an in-app notification with context about what was flagged and a direct path to block, report, or dismiss.

🛑

Block & Report — One Action

Blocking is immediate and silent. The blocked user receives no notification. Reporting submits a structured record for platform review — the reporter's identity is never disclosed to the reported user.

🤝

Wing Companion Support

The Wing AI companion is available throughout the app. If something feels off but you're not sure how to name it, Wing can help you think through what happened and decide what to do — without judgment and without escalating unless you want to.

📋

Repeated Violations → Account Action

Multiple safety flags on a single account trigger escalating review. Confirmed violations result in suspension. Verified serious violations result in permanent removal.

Our Commitment

What we will always do

We will never share your screening responses with other users or third parties. Your psychological profile exists only to power your matches.

We will never notify someone that you blocked them. Blocks are silent and permanent from the blocked user's perspective.

We will always respond to safety reports. Every submitted report is reviewed by a human within 24 hours. If you're in immediate danger, contact local emergency services — no app can substitute for that.

We will improve these systems continuously. AI safety detection is not perfect. When our systems miss something, we want to know. Use the in-app feedback tool or email safety@under-our-wing.polsia.app to report a gap.

A safer place to look for
something real

Join a platform where your safety isn't an afterthought — it's the first thing we built.

Get Started