OpenUp applies AI in clearly defined, limited ways. Our AI features are designed to guide, not to treat. Here's an overview for HR, security, and procurement teams.
AI helps employees reflect, find relevant content, and connect with the right expert. It does not diagnose, give clinical advice, replace an expert, or share conversations with the employer. Every new AI application is reviewed for privacy and safety before launch. Our information security is GDPR compliant and certified to ISO 27001 and NEN 7510.
Read our full description of how we use AI responsibly here: https://openup.com/our-vision-for-responsible-ai-use/
No, never. Organisations only receive anonymised, aggregated insights, never individual conversations that could be linked to an employee.
No. User data is contractually excluded from model training. We do not fine-tune or train any model on user data. Per our agreement with Microsoft Azure, data sent to the API is not used to improve OpenAI models.
Selected anonymised chat transcripts may be reviewed internally for quality improvement of prompts and recommendations during the beta phase, but this is strictly internal and does not involve model training.
All data is currently stored and processed within the EU (Microsoft Azure, Sweden region). We do not transfer user data outside the EEA for AI processing.
The LLM model (GPT-5.3 Chat) is supplied by OpenAI and hosted via Microsoft Azure's EU infrastructure. A Data Processing Agreement (DPA) is in place with Microsoft/Azure in accordance with GDPR Article 28. Under our contractual terms, no customer data is used for model training, not by us, and per our contractual agreements, not by the LLM provider either.
Based on our current assessment, yes. We monitor the phased implementation of the AI Act and assess each AI-supported feature separately.
Yes. An AI disclosure is shown at the start of every conversation. Users are told they are speaking with an AI well-being coach, not a human therapist. The limitations of the AI are stated upfront. This meets the transparency obligations under the EU AI Act for limited-risk AI systems.
The AI well-being coach has built-in emergency detection. When crisis signals (self-harm, suicidality, acute risk) are detected, the AI immediately surfaces a safety message with emergency contact information (112 or local crisis line) and a direct link to 24/7 human support. The AI does not attempt to manage crisis situations itself.
A psychologist. The number is OpenUp’s main phone number, which is staffed as a 24/7 support line. All calls, day and night, are answered by psychologists with a master’s degree in psychology, experience supporting clients, and relevant crisis support training.
The psychologist helps the caller find the right next step on the platform, or, if needed, further help such as specialised crisis services or their general practitioner. In case of immediate danger, the caller is directed to emergency services (112).
Users will be able to manage their consent directly from their MyOpenUp profile settings. Inactive user data is deleted after 36 months of inactivity (defined as no login or usage in 36 months). Active users on active contracts retain all data as long as they keep logging in. Users can request deletion of their data at any time via [email protected]
If you have any questions, our support team is happy to help.