Questions & disclosures

Not yet. The consumer app is not on sale, and there is nothing to download.

Institutional enquiries are open. Schools, organizations, networks and platforms can talk to us now about pilots and deployments.

Yes. The consumer app includes screen time scheduling, app blocking and approval, per-app protection settings, friend-request approval and family location.

Detection is what makes RoseShield different, not what makes it complete. The controls every parent expects are there; what RoseShield adds is reading the interaction itself.

Channel partner agreements are signed in South Africa, Kenya, Uganda, Cameroon and Ghana, with a cooperation framework in the United Arab Emirates.

The first two school pilots are signed in Kampala, and RoseShield Messenger is entering broad pilot testing with a major international children’s organization.

Those are agreements and pilots beginning, not results to report.

Not yet published. Until it is, treat performance claims on this site as our claims rather than verified evidence.

When findings exist, they will be published by harm category, threshold and model version — including where the engine underperformed.

No independent certification today. We will not suggest otherwise, and your technical team will establish it either way.

Expanding the programs already agreed, publishing findings from them, and consumer availability.

No. Their ordinary conversations, friendships and arguments stay theirs, and you never see them.

You are told when something dangerous or inappropriate happens, and you see what you need to make a decision about that specific thing. You do not get a feed, a transcript or a log of their day.

Yes. RoseShield checks a message when your child presses send, so if it was stopped, it happened — and you are told.

What your child gets is the chance to fix it: the message does not reach anyone, they are shown why, and they are offered another way to put it. They learn from the moment; you know about it.

What was found and how serious it is, with the reasoning behind the judgement. Where a decision genuinely requires seeing the content, that is scoped to the item in question — not the surrounding conversation.

Yes, and that is deliberate. Nothing is hidden from them. They can see what it looks at, what it can stop, and what you never get to see. A child who discovers a hidden monitoring tool later loses trust permanently.

On the device. Messages are checked at the moment they are sent. Web addresses and search terms can be checked before a page loads.

Content that does not require a person is analyzed on the device and is not sent to ChildSafe.dev.

When something genuinely needs a person to judge it, the minimum information required for that review goes to you. RoseShield does not create a separate archive of a child's conversations.

No. The engine detects and scores. You decide who is told, whether anything is escalated or reported, and what the situation requires.

In a school or an organization, that decision sits with the designated safeguarding authority, not with us.

Text, images, voice notes and video, recorded and live. It looks for signs of grooming, sextortion, bullying and self-harm, and 34 other categories of harm besides.

Detection runs on the device, so it does not depend on constant connectivity. That is one reason it deploys well in schools with variable infrastructure.

More than a hundred. It follows what a message means rather than checking against a list of banned words, which is why it holds up across slang and when a conversation switches languages mid-sentence.

Performance is tested language by language. Ask us about yours and we will answer honestly.

It will. The question that matters is what a mistake costs.

A mistaken low-risk pause has limited consequences. Your child sees a brief explanation and can rewrite the message.

A mistaken adult alert has much greater consequences, so the threshold for notifying an adult is set much higher.

On lower-risk interventions, yes, and that is a setting you control. As they get older you can let more through, keeping firm limits for serious risks.

Hard limits remain for serious risks at every age.

No, and it will not. A single percentage cannot describe a system whose errors cost different amounts depending on what they trigger. Performance is reported by harm category, threshold and model version.

It does not counsel families. It is detection and decision support, not a support service, and serious alerts carry relevant support resources where available.

It does not report to authorities on your behalf, contact your child's school, or notify another parent. Those decisions are yours.

It does not give you or ChildSafe.dev access to your child’s ordinary conversations.

That depends on the deployment. Built into the handset it covers every application on the device. Inside an app it covers that app. Across a school network it covers the systems the school runs.

With RoseShield Router, every device on the school network is protected — including a student's own phone, for as long as it is connected. Step off the network onto mobile data and it is no longer covered, which is also the edge of the school's duty of care.

In an institutional deployment the customer is the controller and ChildSafe.dev is the processor. The operator agreement comes as a schedule to the contract.

Retention, residency and sub-processor disclosure are set per deployment.

Yes. Ask and we will delete it. We do not sell data, and we do not add enquirers to marketing lists.

No. What constitutes grooming, exploitation or a reportable offense is a matter for the law where you are. The engine detects and scores; your authority decides.