Responsible AI Policy

Last updated: 12 July 2026

This Policy explains, in accordance with the transparency obligations of Regulation (EU) 2024/1689 (the "AI Act") and Articles 13–14 GDPR, which artificial intelligence systems SLP Command uses, what they receive, their limitations, and the human oversight applied to them.

1. AI systems in use

OpenAI — Whisper (transcription)

When you record a Speaking response, your audio is sent to OpenAI's Whisper model for speech-to-text transcription. The resulting transcript is used as input for evaluation.

OpenAI — GPT (evaluation)

Your Speaking transcripts and Writing responses are evaluated by OpenAI GPT models. Evaluation covers:

• Fluency — smoothness, pace, and naturalness of delivery.
• Grammar — accuracy of grammatical structures.
• Vocabulary — range and appropriateness of lexical choices.
• Coherence — logical organisation and clarity of ideas.
• Task Achievement — how fully the response addresses the prompt.

ElevenLabs — Text-to-Speech

Listening exercises and pronunciation models are generated using ElevenLabs text-to-speech technology, from exercise text only. No personal data is sent to ElevenLabs.

Adaptive Coach and Intelligence Dashboard

These features are not powered by external AI inference. They are deterministic statistical computations performed on your own historical results within our own infrastructure, to suggest what to practise next.

2. What each AI system receives

• Speaking: your audio recording and the exercise prompt text.
• Writing: your written response and the task prompt.
• Listening generation: exercise text only — no personal data.

We do not send your name, email address, account identifier, or any other directly identifying information to an AI provider together with your Content.

3. AI Act risk classification statement

SLP Command has assessed whether its AI-based evaluation features fall within Annex III of the AI Act (high-risk AI systems used for the evaluation of learning outcomes).

Our conclusion, reviewed as of the date of this Policy, is that the Service does not constitute a high-risk AI system under Annex III, because:

• all AI Feedback is explicitly, consistently, and prominently disclaimed throughout the Service as indicative, non-binding, educational guidance, never as an Official Assessment;
• no admission, certification, employment, promotion, or other legally or similarly significant decision about you is made, wholly or partly, based on AI Feedback generated by the Service;
• the Service is used exclusively for individual, self-directed study, and is not integrated into any institutional admission, evaluation, or certification process.

This classification is reviewed whenever the Service's use context changes materially — in particular, before any institutional, governmental, or academy-facing deployment in which AI Feedback could feed into a real admission, certification, or personnel decision. Should such a deployment occur, SLP Command commits to re-conducting this assessment and, if the high-risk threshold is met, implementing the corresponding obligations under Articles 9–15 of the AI Act before that deployment goes live.

4. Transparency obligations (Article 50 AI Act)

You are informed, both in this Policy and at the point of interaction within the Service, that you are interacting with an AI system when receiving Speaking or Writing feedback. AI-generated transcripts and evaluations are clearly labelled as such within the Service interface.

5. Important limitations

• AI feedback is indicative guidance, not an official SLP / STANAG 6001 assessment.
• AI models may produce inaccurate, inconsistent, or incomplete evaluations.
• Scores and estimated levels shown in the App do not predict or guarantee any outcome in an official exam.
• AI systems reflect the biases present in their training data. Always consult official examiners and materials for real exam preparation decisions.

6. Human oversight

SLP Command reviews aggregate AI Feedback quality on an ongoing basis and maintains a "Report a Problem" channel through which any user may flag inaccurate or inappropriate AI output for human review. No automated system within the Service takes any action against a user's account, access, or standing based solely on AI Feedback content.

7. Data and your rights

Your audio and text are processed by AI providers acting as sub-processors under data processing agreements. For full details on retention, deletion, and your GDPR rights, see the Privacy Policy and the Subprocessors page.

8. Contact

To report an AI-related error or concern: support@slpcommand.com. For AI Act or governance queries: legal@slpcommand.com.