Responsible use

​

Tell people it is AI

When your app speaks with a Namu voice, let listeners know the voice is generated, the first time they hear it and wherever it matters: health, money, official or legal information. A short line is enough: “This message is read by a computer voice.”

​

Ask before you listen

Recordings sent to /v1/transcribe, /v1/translate and /v1/converse are people’s voices and words. Translation and the voice agent pass the text they work on to Google Gemini through OpenRouter. Tell people their voice is sent to a service to be understood, get their agreement, and do not send recordings of people who did not agree, such as voice notes forwarded from others.

​

Do not use it to deceive or harm

Do not use the API to:

• impersonate a real person, an official or an organisation, or make people believe they hear someone they do not;
• defraud, for example fake calls or messages about mobile money, prizes or debts;
• mislead people about elections, public health or emergencies;
• harass, threaten or shame anyone;
• make decisions about people (health, credit, jobs, justice) from a transcript or translation without a person checking it.

​

Keep a person where mistakes hurt

Speech recognition and translation make mistakes, more often with noise, names and numbers. Where a mistake could hurt someone, show the text back, ask for confirmation, or have a person check.

​

Respect the languages

Have native speakers listen to the audio and read the transcripts your app produces before launch. Hausa, Zarma and French speakers will notice what a test script cannot, and they will tell you, and us, how to do better.

​

Report misuse

If you see the API used to deceive or harm, write to contact@namuai.org.