AI-Generated Content Notice
This notice explains which Read OutWard features use artificial intelligence (AI), how we label AI-generated content, and the limits of that content.
Features that use AI#
| Feature | What AI does | Provided by |
|---|---|---|
| Narration and tap-to-hear | Generates English narration for articles, books, interactive books and your imports with AI voices | ElevenLabs; MiniMax (through fal.ai) |
| Voice portraits | The portraits in the voice picker are AI-generated fictional people, not the voice actors | AI image generation |
| Illustrations and covers | Some article, book and interactive-book illustrations are AI-generated | AI image generation |
| Editorial content | Some Chinese translations, sentence notes, word definitions, quiz items and news summaries are produced with AI assistance and checked in our editorial process before publication | Large language models |
| AI explanations and translations (members) | Generates definitions, translations or explanations for sentences you select (including subtitles of episodes you imported) | OpenAI models (through fal.ai and OpenRouter) |
| Web scene conversations | Turns your speech into text and replies in character | Speech-to-text and large language models (through fal.ai) |
| Chinese translation of imported articles (members) | Machine translation done on your device | Apple system translation |
| Pronunciation scoring | Automatically assesses pronunciation, fluency and completeness | iFLYTEK speech evaluation |
Which data each feature sends and to whom is described in the Privacy Policy and Third-party services.
How we label AI content#
- The voice picker, interactive-book details and the interactive-book player are labelled “AI voice”, “AI-generated portrait” and “some voices and illustrations are AI-generated”, and name the voice provider.
- Chinese translations of imported articles are labelled “automatic translation”.
- This page is the overall notice for these features; help pages in the app and on the website link here.
About iRe#
iRe is Read OutWard’s study-plan coach, not a chatbot. It takes no free-text input; its reminders, suggestions and encouragement come from fixed templates filled in with your learning data on your device, without calling a large language model.
iRe does not study for you and never tells you an answer before you have tried. In quizzes and exercises you see the correct answer and explanation only after you answer; hints in interactive books come in steps and do not give the answer away. Some exercises have a Show answer button that reveals it only when you tap it.
Accuracy and limits#
- AI voices may mispronounce some words or get stress and intonation wrong; AI definitions, translations and explanations may be inaccurate, incomplete or miss the context; characters in scene conversations may say things that are not factual.
- Pronunciation scores are automatic assessments that can be affected by your recording environment, accent and microphone. They are practice feedback only and do not reflect any exam’s marking.
- AI content is for language learning only and is not professional advice (for example legal, medical or financial). Do not rely on it for important decisions.
- We keep reviewing and improving AI content. If you find an error or inappropriate content, report it in the app or on the website, or email support@readoutward.com, and we will act promptly.
AI credits#
You spend AI credits only when narration is newly generated for you, for example for content you imported or in a voice we have not pre-generated. Pre-generated narration, cache hits and failed generations cost nothing; AI explanations and translations are included in membership and cost no credits.
Monthly grants: free accounts 20, trial 50, members 300, not rolled over; you can also buy credit packs that never expire. See the Subscription Terms for the rules.
Your content and AI#
What you send to AI features (such as selected sentences, imported text or your replies in a scene conversation) is used only to produce the result you asked for. We do not use your content to train models of our own. Providers process it under their business terms. When using AI features, follow the acceptable-use rules in the Terms of Service and do not enter other people’s personal or sensitive information.