Version 0.1 · 2026-09-23Draft document
This English version is provided for convenience only. The Kazakh and Russian versions are legally binding and have equal force. In case of any discrepancy, the Kazakh and Russian versions prevail.
Notice on the Use of Artificial Intelligence
The Көпір platform uses artificial intelligence systems. This Notice takes into account the requirements on transparency of the use of artificial intelligence systems established by Law of the Republic of Kazakhstan No. 230-VIII of 17 November 2025 "On artificial intelligence".
1. Where AI is used
| Feature | What AI does | Who decides |
|---|---|---|
| Task clarification | Finds what information is missing and asks at least three questions | The business answers the questions |
| Task card | Arranges the description and answers into the fields of the card | The business reviews, edits and confirms every field |
| Technical documentation | Prepares a description of the task for students | The business edits and confirms it |
| Recommendations | Explains why a task fits a team | The team chooses where to send a proposal |
| Problem discovery (Discover) | Finds recurring problems in connected sources and cites supporting quotes and numbers | The business decides whether to turn a finding into a task |
| Meeting transcription | Turns a meeting recording into text | The business reviews the text |
AI does not calculate the readiness rating. It is calculated by software under the open formula of the Catalog and Rating Rules.
2. Rules we follow
- AI does not add information the User has not provided. If there is no information, the field stays empty.
- Every finding of AI in the Discover section is backed by a quote or a number from the source. Findings without support are discarded.
- Everything AI prepares is reviewed and confirmed by a person before publishing.
- AI does not choose a team for the business and does not assign contractors.
- Recommendations do not restrict the catalog. All tasks are available to all teams.
- AI does not use personal or sensitive characteristics of participants.
3. Labeling
Text prepared by AI is marked on the platform until a person confirms it. Such text is shown with the note "Prepared with AI. Review and confirm before publishing."
4. Limitations
4.1. AI answers may be inaccurate or incomplete. They are auxiliary.
4.2. The User assesses the result of AI and is responsible for the decisions made on its basis.
4.3. If the AI service is unavailable, the platform works with pre-recorded sample answers. The interface says so.
5. User rights
The User has the right to:
- know which features of the platform work with AI;
- correct or reject any result of AI;
- not use AI features and fill in the card manually;
- ask for an explanation of a result and demand that it be reviewed by an employee of the Operator by writing to будет указано при запуске.
6. Data
6.1. For AI to work, texts are transferred to the providers OpenAI, NVIDIA. Data processing is described in the Personal Data Policy.
6.2. Before sending, the platform looks for personal data in the text and offers to depersonalize it.
7. Obligations of the Operator
The Operator ensures transparency of AI, labels its results, checks the quality of answers on test examples and suspends a feature if a material error is found.