Table of Contents
Transparency in the use of AI
Tallinn University of Technology (TalTech) uses artificial intelligence (AI) in teaching, research, work, and communication processes. When using AI, we comply with applicable legislation, including the European Union Artificial Intelligence Act (AI Act), and consider transparency and human oversight important.
When do we disclose the use of AI?
We disclose the use of AI when:
- a person interacts directly with an AI chatbot, AI agent, or other AI system;
- we publish an image, audio, or video created or significantly modified using AI that may give the impression of a real person, event, or situation;
- we publish AI-generated or AI-modified text that concerns a matter of public interest and whose content has not been substantively reviewed by a human.
If the use of AI is already clearly apparent to a person based on the context, a separate disclosure may not be necessary.
A deepfake is an image, audio, or video content created or significantly modified using AI that may give a misleading impression of a real person, event, or situation. Such content must be clearly labelled.
For example:
- a video in which a real person appears to say something they did not actually say;
- an artificially generated or modified voice that imitates a real person's voice;
- a photo depicting a person at an event they did not actually attend;
- an image or video depicting an event that did not actually take place.
Not all visuals created or modified using AI are deepfakes. For example, a stylized illustration or a clearly fictional character may not give the impression that something actually happened in reality.
If AI-generated or AI-modified text concerns a matter of public interest, we assess before publication whether the content has been substantively reviewed by a human.
A substantive human review includes, among other things, that a person:
- verifies the accuracy of the content and, where necessary, checks the sources;
- assesses the meaning and context of the text;
- makes the necessary corrections;
- decides whether the content is suitable for publication.
Proofreading or correcting spelling and grammar alone does not constitute a substantive human review.
AI may be used, for example, to generate ideas, structure text, translate, create summaries, or provide other forms of assistance. In such cases, the use of AI does not automatically mean that the final result must be separately labelled.
A human who reviews and approves content for publication is always responsible for the published content.
Information related to the use of AI
Data protection and information security
When using AI tools, TalTech's data protection and information security requirements must be followed. Personal data, confidential information, trade secrets, or other protected information must not be entered into AI tools unless there is an appropriate legal basis and authorization to do so.
TalTech's principles for processing personal data, data subject rights, and information security measures are described in the TalTech privacy policy.
Legal Basis and Further Information
AI transparency obligations arise from Article 50 of the European Union Artificial Intelligence Act (AI Act). The obligations apply from 2 August 2026.
Further information is available from: