AI Transparency Notice — FizzUp

1. Key notice

FizzUp uses artificial intelligence systems. Users should always know when they are interacting with AI rather than a human. AI can produce incorrect, incomplete or contextually inappropriate content.

2. How AI is used

AI supports, among other things:

• text and voice conversations;
• language analysis and feedback;
• personalized exercises and next steps;
• generation of scenarios, questions and examples;
• summaries and educational materials;
• speech processing and transcription.

FizzUp may use models and services from different providers. A provider change does not change the fact that the user is using the FizzUp service.

3. AI providers

The provider registry describes admitted integrations and conditions of use. The recipient depends on the feature and its active configuration; listing an integration does not mean that it receives data. We do not declare one common provider for every feature.

Before transferring data, the feature identifies the AI provider selected for that operation, the data categories and processing purpose, and links to the current provider registry. For a live conversation, the notice covers audio and transcript. Acknowledging the notice is separate from accepting the Terms; it is not described as consent where processing is based on performing the user-requested service. Without the required acknowledgement, the feature does not transfer data. If a provider change, including a fallback, requires transferring data to another recipient, the notice and required acknowledgement are updated before that transfer. If the recipient cannot be resolved or is not admitted, the operation does not start.

4. Data and model training

FizzUp does not use private conversations, transcripts, recordings or coaching memory to train general AI models.

FizzUp configures external services so that, under available contractual terms and provider settings, transferred data are not used to train the provider's general/public models.

Anonymous or effectively aggregated data may be used to improve teaching methodology, assess exercise effectiveness and develop the product.

5. Limitations

AI output is not guaranteed to be correct. FizzUp should not be used as the sole basis for:

• an employment decision about another person;
• educational admission or an official exam decision;
• medical or psychological diagnosis;
• legal, investment or credit advice;
• life-or-health critical decisions.

A FizzUp language-level result is educational feedback, not an official certificate.

6. B2B and no employee scoring

In organization plans, an employer does not receive individual transcripts, errors, scores, goals or coaching profiles. FizzUp is not an employee-ranking system.

Organization outcome/progress reporting is aggregated and hidden for groups smaller than 10 active participants. By name, the organization sees only invitation, account activation and onboarding completion.

7. Voice and synthetic speech

The conversation voice may be synthetic. A feature may process user audio in real time to recognize speech, understand an utterance and generate a response.

The interface clearly indicates that the user is interacting with AI and that the voice may be synthetic. Where FizzUp is a provider of a system generating synthetic content within the meaning of the AI Act, it applies the required machine-readable technical marking of outputs, taking applicable transitional provisions into account.

Live conversation audio is not stored by FizzUp. The transcript text is stored (90 days by default). The only voice recordings kept are comparison-task recordings — only when the user turns saving on, and for up to 7 days.

8. Human control

The user controls whether to use AI features, may stop a conversation, withdraw optional consent where consent is the processing basis, delete history and report a problem. FizzUp monitors system safety and may change configuration, restrict features or change providers where required for security or legal compliance.

9. AI governance

DB 77 CUW maintains an internal AI-system/provider register, performs risk assessment, maintains AI-governance policies, reviews material changes and provides AI-literacy training to relevant staff.

AI questions: support@fizzup.ai. Privacy questions: privacy@fizzup.ai.