Skip to content

Poodll AI Data Map

This is a guide to Poodll’s AI services, notably how they process student data, by region and by plugin. It is designed to help you configure Poodll to be compliant with the data policies and regulations that apply in your organisation.

Download this guide as a PDF


Schools ask three main things about Poodll and AI:

  • Service Providers. What AI services does Poodll use? And who provides them?
  • Location. Where is student data processed?
  • Training. Is student data used to train anyone’s AI models?

This guide answers those questions.

  • Your region. Every Poodll plugin has an AWS Region setting (Frankfurt, Dublin, Tokyo, Sydney and so on). Recordings and transcripts are stored in that region. Poodll chooses AI providers with endpoints (i.e. servers) appropriate for that region. Each Poodll plugin is independent of the others. So the region needs to be set in each plugin.
  • The student data we process. Audio and video recordings, speech transcripts and student submitted text. Student names, emails, grades or other personal data are not submitted.
  • Data retention. After processing, Poodll stores only recordings and transcripts. Each plugin has a default Days to keep file setting. After this time those recordings and transcripts are deleted from storage. Not all recordings are stored or transcribed.
  • Local AI. Since Moodle 4.5, Moodle has its own AI settings (Site administration > General > AI). Some Poodll features can use the AI provider you set up there instead of ours. That could be a provider in your country, or a model you host yourself. The information tables below call this Local.
Kind of AIWhat it doesDoes it process student data?
TTS (text to speech)Computer voices that read lesson text aloudNo. It reads lesson text written by the teacher
STT (speech to text)Writes down what a student said, and scores pronunciationYes, the student’s voice
Audio ChatA live spoken conversation between the student and an AIYes, the student’s voice and the conversation
AI EvaluationMarks a student’s speaking or writing and gives feedbackYes, the student’s words, as text
Image GenerationMakes pictures for lessonsNo. It uses the teacher’s description
Text Content GenerationWrites lesson content: passages, questions, definitions, translationsNo. It uses teacher provided prompts, not student data.

World map of the Poodll regions. Blue pins mark most regions: N. Virginia, Ottawa, São Paulo, Tokyo, Singapore, Sydney, Mumbai and Bahrain. Green pins mark Europe and South Africa: Dublin, London, Frankfurt and Cape Town. A red pin marks Ningxia in China.

PackagePluginTTSSTTAudio ChatAI EvaluationImage GenerationText Content Generation
Poodll MediaCloud Poodll Assignment (submission)—✅————
Cloud Poodll Feedback—✅—✅——
Cloud Poodll Question—✅————
Tiny Poodll—✅————
Poodll Filter✅—————
Poodll LanguagesReadAloud✅✅—✅—✅
Solo✅✅—✅—✅
WordCards✅✅——✅✅
MiniLesson✅✅✅✅✅✅
RegionTTSSTTAudio ChatAI EvaluationImage GenerationText Content Generation
Most regions
US East, Ottawa,
São Paulo, Tokyo, Singapore,
Sydney, Mumbai, Bahrain¹
AWS(US)
Google(US)
Azure(US)
LemonFox(US)
OpenAI(US)
AWS(your)²
Poodll(your)²
AssemblyAI(US)
LemonFox(US)
Azure(US)
Google(US)

Optional: OpenAI(US)⁴
OpenAI(US)

Optional: Local³
OpenAI(US)
Google(US)
Optional: Local³
OpenAI(US)

Optional: Local³
Europe and South Africa
Dublin, London, Frankfurt,
Cape Town
AWS(US)
Google(US)
Azure(US)
LemonFox(US)
OpenAI(US)
AWS(your)²
Poodll(your)²
AssemblyAI(EU)
LemonFox(EU)
Azure(EU)
Google(US)

Optional: OpenAI(EU)⁴
OpenAI(EU)

Optional: Local³
OpenAI(EU)

Optional: Local³
OpenAI(EU)

Optional: Local³
China
Ningxia
AWS(China)AWS(China)
Poodll(China)
Not availableDeepSeek(China)

Optional: Local³
AWS(China)

Optional: Local³
DeepSeek(China)

Optional: Local³
  1. Bahrain cannot be reached at the time of writing (19 November 2026), after damage in the Iran–US conflict.
  2. Your means “Your Region” the AWS region chosen in the plugin’s settings. Poodll STT and AWS STT run in ‘your region.’
  3. Local means one of the AI providers registered in your Moodle site. Only some features can use it.
  4. Audio Chat. To use OpenAI you must provide your own OpenAI API key, and pay charges. Poodll’s Europe region OpenAI will only work with a key from an OpenAI project whose data residency is set to Europe.

In most cases your student data will not be used for training. But you may need to disable browser STT.

ProviderUsed for training?Their policy
Amazon Web ServicesNo. Poodll has opted out for all AWS AI servicesAWS opt-out policy
Poodll’s own speech engineNo.n/a
AssemblyAINo. EU servers are never used for training, and Poodll has opted out for the US serversData retention and training
LemonFoxNo. Policy states “All data is deleted immediately after processing.” Inference only.Privacy
OpenAINo. API data is not used for training. It is kept for up to 30 days for abuse checks, then deleted. For Dublin, London, Frankfurt and Cape Town that copy is also kept in the EUYour data
Google (Gemini)No. On commercial plans Google does not use prompts or responses to improve its productsGemini API terms
Google CloudNo. TTS only.Google Cloud privacy notice
Microsoft Azure SpeechNo. Audio is scored in memory and not storedSpeech to text data and privacy
DeepSeek (China)Yes. Their policy allows training usePrivacy policy
Local (your provider)Provider dependent. Check the settings and terms of your agreement with them.n/a
Browser STTProbably. Turn it off if necessary. (see below)n/a

Poodll activities that deliver instant transcription may use the speech recognition built into the student’s browser. In Chrome the audio goes to Google, in Safari to Apple, and in Edge to Microsoft under their consumer terms. Poodll has no service agreement to cover browser STT and cannot configure a region or data usage terms. To be sure that student data is not used to train AI models, browser STT should be disabled in the Poodll plugins that use it. This will make transcription slower in some cases and may affect the quality of the transcription (for better or for worse).

PluginTurning off browser STT
ReadAloudThe practice step can use browser STT. To disable it, in the plugin’s admin settings set “Open STT realtime” to “3rd party” (not Auto). Alternatively choose “Guided STT (Poodll)” as the “Practice step transcriber”.
SoloTo disable browser STT, in the plugin’s admin settings set “Streaming recorder by default” to “No, use the standard recorder” or “Yes, use cloud recognition”.
WordCardsThe speechcards item can use browser STT. Set “Transcriber” to “Guided STT (Poodll)” to use Poodll STT instead. Alternatively do not use speechcards.
MiniLessonTo disable browser STT in the MiniLesson items that use it, in the plugin’s General Settings admin page, check “Enable Alternate streaming”.

PluginWhat else you should know
Cloud Poodll Assignment SubmissionRecordings are stored in your region. Speech to text is used if Enable Transcription is on, to make a transcript of the student’s recording. Switch it off and no AI is used
Cloud Poodll Assignment FeedbackSpeech to text is used if transcription is enabled. AI Evaluation is used only when the teacher uses corrections feedback and clicks the AI corrections buttons.
Cloud Poodll QuestionSpeech to text is used if transcription is enabled. Switch it off and no AI is used
Tiny PoodllUses speech to text to make subtitles for recordings. Subtitles are optional and can be switched off
Poodll FilterSome widgets can use TTS to read page text aloud.
ReadAloudThe Read and Practice steps use speech to text. Word-by-word comparisons with the passage are done on your Moodle server, and no generative AI is used. AI content generation is available for teachers to create reading passages.
SoloAI Evaluation gives an AI grade, feedback, grammar correction and a level estimate. It is switched on or off with the activity’s Enable AI setting. With it off, Solo still records, transcribes and counts words but there is no automated evaluation.
WordCardsSpeechcards practice activity uses speech to text. Image and text generation are teacher tools that make pictures, definitions and translations from the word list. No student data is sent. Translation and pictures can use Local AI.
MiniLessonEach lesson can contain different items, and the AI used depends on the item type. Image and text generation for building lessons and items are teacher tools. Any item type can be switched off for the whole site in Manage Lesson Items. Per item details are in the table below.

Each item has its own AI usage profile, and some items do not use AI at all.

ItemStudent data sent?AI usedWhat else you should know
Audio ChatYes. Audio and text transcriptAudio Chat, AI EvaluationThe conversation runs on Google Gemini by default, which gives no region choice. To keep it in Europe you need to provide an OpenAI API key. Create a project in your own OpenAI account with Europe as its data residency, make an API key in it, enter it as the OpenAI Key under MiniLesson’s Other API Keys (BYOK), and set the item’s Provider to OpenAI. OpenAI then bills you directly. The data residency of the project that generates the key must be Europe.
CardsNoTTSVocabulary cards with definitions and images.
DictationNo.TTSListen and type out the sentences.
Dictation ChatNo.TTSListen and type the missing words.
FictionNoTTSAn interactive story. Story text can be read aloud and translated
FluencyYes. Audio.STTPronunciation scoring with STT by Microsoft Azure: West Europe for Dublin, London, Frankfurt and Cape Town, US East for most other regions. The Transcriber setting does not change this.
Free SpeakingYes. Audio and text transcriptSTT, AI EvaluationMarks a spoken answer and gives feedback. AI Evaluation can use Local AI: in AI Manager, set Grade student submission to an AI provider set up in Moodle
Free WritingYes. The student’s writingAI EvaluationMarks a written answer and gives feedback. Can use Local AI, as for Free Speaking
H5PNoNoneRuns an H5P activity.
Listen and RepeatYes. AudioTTS, STTThe student’s words are checked against the expected words. No generative AI
Listening Gap FillNoTTSListen and type the missing words.
MC Audio (Multi Choice Audio)Yes. AudioTTS, STTMulti choice question where the student speaks the answer. Checked against the expected words. No generative AI
MultichoiceNoTTSMultichoice question
Multichoice QuizNoTTSMultiple multichoice questions
PageNoTTSA content page which is not graded.
Passage Gap FillNoTTSListen and type the missing words.
Passage ReadingYes. AudioSTTThe student reads a passage aloud and is scored by comparing words, as in ReadAloud. No generative AI
ScatterNoTTSMatch the cards.
Self CheckNoTTSThe student records, listens back and judges their own attempt. No speech recognition
ShadowNoTTSThe student speaks along with a YouTube clip. No recording and no speech recognition
Short AnswerYes. AudioTTS, STTChecked against the expected answers. No generative AI
SlidesNoTTSSlideshow presentation.
Smart FrameNoNoneShows an html page/widget uploaded by the teacher.
Space GameNoTTSMatching activity that works as a shooter game.
Speaking Gap FillYes. AudioTTS, STTChecked against the expected words. No generative AI
Speech CardsYes. AudioTTS, STTChecked against the expected words. No generative AI
Speed ReadingNoNoneTimed reading and comprehension check
Typing Gap FillNoTTSFill in the blanks type activity
Word CardsNoTTSVocabulary practice activities
Word ShuffleNoTTSArrange the words in the correct order