AI Use Policy

AI Use Policy

Part A. For learners and parents (public)

A1. What AI does on Edly. (1) An AI study assistant answers questions about the lesson you are on. (2) AI can generate hints for quiz questions. (3) Mentors may use AI to draft quiz questions, rewrite lesson text, transcribe videos, or create lesson audio or avatar video. (4) AI may help organise course structure for Mentors, who remain responsible for the final content.

A2. Provider. Our AI features currently use third-party AI services (currently OpenAI; speech, avatar and transcription services are also used for production). The text you type, and the lesson text needed for context, is sent to the provider. See the Privacy Policy for transfer safeguards.

A3. What we send and what we don't. We send your question, recent chat history and the relevant lesson excerpt. We do not intentionally send your name, e-mail, phone number or payment data. Please do not type personal, financial, health or confidential information into an AI chat. AI chat content is stored and may be reviewed by a person to handle safety flags, abuse and quality; the retention period is in the Privacy Policy.

A4. It can be wrong. AI answers may be inaccurate, incomplete or out of date, and every AI answer is marked as possibly inaccurate. It is a study aid, not an expert, tutor of record, or adviser (professional, legal, medical, financial). Check important points with your Mentor or an authoritative source. Do not use it as the sole basis for exam answers or decisions about work.

A5. No personal decisions about you. AI does not decide your grade, certificate, refund or account status. Those are decided by rules and people.

A6. Safety for children. The assistant refuses topics unrelated to the lesson, is instructed not to reveal system instructions, and flagged exchanges are escalated to a person. Parents can ask for deletion of a minor's chat history.

A7. Your responsibility. Do not use AI to cheat in assessed tasks, create deepfakes of real people, harassment, or unlawful content. Follow the AI tool provider's terms when you use external tools taught in courses.

A8. Rate limits and fair use. Per-user and global limits apply. Abuse (automation, extraction of system prompts, attempts to bypass safety) may lead to suspension. Do not post or share AI-assistant answers that reproduce paid lesson content (Terms of Use, clause 5.4).

A9. Your rights. You may ask for access to or deletion of your AI chat history and object to its review, subject to safety and legal needs.

Part B. Rules for mentors and staff (binding)

B1. Responsibility. AI-assisted content is the Mentor's content. The Mentor must review every AI-generated line, fact, formula, number and example before publishing and remains responsible for accuracy, originality and compliance with the Content Standards.

B2. No confidential or personal data in prompts. Never put learner data, employer or client confidential material, trade secrets or other people's personal data into external AI tools.

B3. Rights and licences. Use only AI tools and outputs whose terms allow commercial use and sublicensing to Edly. Keep a record of the tool and plan used for each Course asset. Do not use tools whose terms prohibit the use or require attribution you cannot give.

B4. Training material. Do not use copyrighted material you do not have the right to use as input to generate lessons; do not imitate the style, voice or likeness of a real person without a written release.

B5. Disclosure. Disclose AI-generated voices, avatars and synthetic imagery to learners where it could mislead (a short label on the lesson).

B6. Staff. Staff may use approved AI tools only for approved purposes. No customer or learner data in non-approved tools. Log high-impact use. Review outputs before they leave Edly.

B7. Prompt and key security. API keys are held in the secret store, not in code or chat. Rate limits and monthly spend caps are set at the provider. Provider and model changes go through change management and are reflected in the sub-processor list.

B8. Incident handling. Harmful, biased or privacy-leaking outputs are logged, the prompt fixed, and—if personal data is involved—handled under the Incident Response Plan.

B9. Review. This policy is reviewed every 6 months and when the provider, model or law changes.

B10. Edly-generated output stays on Edly. Output created with Edly's own generation features (AI course builder, lesson and quiz generation and rewriting, transcription, text-to-speech, avatar and video rendering) may be used only as part of the Course on the Platform, and not on other platforms or channels (Author's Licence Agreement, clause 4.7).