Safety

Last updated: 7 August 2026

1. If you need help right now

These services are staffed by trained people. They are free, confidential and available whether or not you are in crisis — you do not have to be at the worst point to call.

If someone else is at risk and you are not sure what to do, the same numbers will talk to you about how to help them.

2. What Filomou is, and what it is not

Filomou is a language-learning app. Its buddies are AI characters with AI-generated voices. They are designed to feel like a friend abroad who keeps in touch, because that framing makes people practise more often than a lesson does. That design choice makes the next sentence important.

• A buddy is software. Any warmth, memory, concern or relationship it expresses is generated by a language model. It is not a person and it does not know you in the way a person would.
• It is not therapy, counselling or medical care, and it is not a substitute for a doctor, a therapist, a crisis worker or a human teacher.
• It is not an emergency service. A buddy cannot call an ambulance, contact your family, alert anyone, or reach the outside world in any way. It cannot act. It can only talk.
• Nobody is listening live. We do not monitor calls or chats in real time, and no human is standing by to intervene.

3. What happens if you tell a buddy you are struggling

We maintain a protocol for conversations that touch suicide, self-harm or serious distress. This section describes what that protocol actually does, so you know what to expect before it happens.

1. The character steps aside. Every buddy is instructed that if you express thoughts of suicide, self-harm, or that you are in danger, the language lesson stops and the roleplay stops. It will not stay in character, it will not treat it as practice material, and it will not carry on with the scenario.
2. It responds plainly and kindly, in a language you understand. Filomou normally pushes you to speak the language you are learning. In these moments it switches to your own language, because being made to struggle for words is the last thing anyone needs then.
3. It points you to real help. The buddy names the crisis resources for your region — the ones in section 1 — and encourages you to contact them or someone you trust.
4. It will not do certain things, ever. Buddies are instructed never to provide methods, means, encouragement, instructions or “how-to” information relating to suicide or self-harm; never to romanticise, dramatise or roleplay such an act; never to discourage you from seeking help; and never to claim to be a therapist, a counsellor, a doctor or a human being.
5. It does not attempt to treat you. A buddy will not try to provide therapy, diagnose you, or take on a clinical role it is not capable of filling. Its job in that moment is to be kind, be honest about what it is, and get you to someone who can actually help.

We also apply the safety filters of our AI provider, Google, on top of our own instructions.

4. What we cannot promise

We would rather state this plainly than imply a level of protection we cannot deliver.

• Language models are not deterministic. Safety instructions substantially change how a model behaves, but they cannot guarantee that every response in every language and every context will be right.
• We do not detect risk that is not expressed in the conversation.
• We cannot contact you, your family, or emergency services on your behalf, and we will not silently escalate a private conversation to a third party.
• A buddy cannot tell whether you mean something literally or are practising a phrase. It is instructed to err on the side of taking you seriously, which means it will sometimes break character when you did not need it to. We accept that trade-off deliberately.

5. Adults only

Filomou is for people aged 18 and over. It is not offered to minors in any country, even with a parent's permission. This is a deliberate decision: a proactive AI character that phones you, remembers you across calls and is built to feel like a friendship is not something we are willing to put in front of a child.

We enforce this by requiring an explicit adult confirmation before any AI feature is enabled, by recording that confirmation on the account, by configuring both app stores for an adult audience, and by acting on the age signals the app stores provide where they are available. We keep our marketing and store listings adult-directed.

If you believe someone under 18 has created an account, write to privacy@filomou.com. We will investigate, and where we confirm the account belongs to a minor we will disable it and delete the associated personal data within 30 days.

6. Reporting a safety problem

If a buddy says something harmful, unsafe or inappropriate, please tell us. We read every report and use them to correct the instructions our characters run on.

• Safety and content reports: support@filomou.com
• Privacy, data and underage accounts: privacy@filomou.com

Include roughly when the conversation happened and what was said, as far as you are comfortable sharing. You never have to send us anything you would rather keep private.

7. How this protocol is maintained

The safety instructions above are built into the system prompts that every buddy, lesson and practice scenario runs on, so they apply across voice calls, chat and every character in the app. We review them:

• when we change the AI model or provider;
• when we add a language, a character or a new conversation surface;
• when a user report or our own testing shows a gap;
• and at least once every twelve months regardless.

This page is updated whenever the protocol changes, and the date at the top reflects the current version.

8. Contact

Adam Daghmah, trading as Dev.Enkrio (KVK 42061187; VAT ID NL005465302B85)
Brouwhuis 52, 1695 JS Blokker, the Netherlands
support@filomou.com