1. Scope of application
This addendum supplements the contractor's general terms and conditions (GTC) for all services related to artificial intelligence (AI), in particular:

• development and integration of AI applications (e.g. based on Brain.js, transformer models, etc.),
• use of external AI services via APIs (e.g. OpenAI, Anthropic, Mistral, etc.),
• Creation of training data, tokenization, data preparation and evaluation.
  In case of contradictions between the GTC and this addendum, the provisions of this addendum shall take precedence for AI-related services.

2. General technical and legal conditions.
The contractor only offers practical, lean and task-oriented AI solutions. Large-scale language models such as ChatGPT, Claude or comparable systems are not part of our own AI developments due to their complexity, infrastructure dependency and license conditions.
The integration of such services is only possible via officially available APIs and requires a valid license from the client.

3. Data processing and data protection
The contractor is obliged to comply with the applicable data protection regulations (in particular the GDPR), insofar as these are within the contractor's sphere of influence.
If external AI services are integrated via interfaces, the responsibility for their data protection-compliant use lies with the client, especially in the case of personal or confidential information.
The client is obliged not to transmit any data whose use or processing violates legal or contractual provisions.

4. results and model behavior
For machine learning-based applications – in particular for generative AI models or transformer-based systems – no guarantee can be given for certain results or their accuracy.
The generated content, answers or decisions are based on probabilities and training data. The contractor assumes no responsibility for the legal, substantive or factual correctness of the output generated by the model.

5. Performance limits and responsibility
The AI solutions developed as part of a project are limited to the agreed use case and scope. Any further use, modification or repurposing of the AI models outside the originally defined purpose is at the client's own risk.
The contractor is not liable for damages resulting from the independent or improper use, disclosure or modification of the AI components.

6. Further development & maintenance
AI models and their training data can become outdated or prone to errors over time. Automatic further development, maintenance or retraining is not part of the service, unless this has been explicitly contractually agreed.

7. final provisions
This addendum is an integral part of the respective contract for AI-related services. It automatically comes into force upon commissioning.