Security measures

Last updated:21.07.2025

At Praven Intelekt, the security and confidentiality of your data are fundamental to us. We work to protect your information at every stage of your interaction with our platform, applying technical and organizational measures in line with applicable law, including the General Data Protection Regulation (GDPR).

Our security measures at a glance:

• All data is stored and processed exclusively within the European Union, on servers located in Germany, France, and Sweden.
• The data you upload to the platform is not used to train Praven Intelekt’s language models.
• Your data is not shared with OpenAI or any other third-party providers, and it is not used to train their models.
• We use modern encryption of data (including AES-256) both at rest and in transit, together with trusted cloud services certified to ISO 27001 and other international standards.

If you have questions about our security policy or would like more information, you can reach us at [email protected].

Hosting and data storage

We use leading cloud providers with a high level of security to store and process data in geographically compliant jurisdictions.

EU hosting and data storage: All data uploaded to Praven Intelekt is stored securely within the European Union, in compliance with the GDPR. Data is hosted on AWS servers (Frankfurt), while the compute models run through Azure data centers in France and Sweden. For our models, we use the dedicated Azure OpenAI Service, which guarantees that no customer data is shared with OpenAI or transferred outside the EU. You can read more about the security measures of the Azure OpenAI Service here.

Enterprise B2B solutions: For enterprise clients, we offer separate, isolated instances of our platform. Data is stored and processed exclusively within that infrastructure and is never shared between clients.

Technical security measures

Encryption: Our platform applies up-to-date encryption and security measures both in transit and at rest:

• All communication between your browser and our servers is protected via the SSL/TLS protocol.
• Data is encrypted with AES-256, one of the most secure encryption standards.
• We offer the option of double encryption with customer-managed keys for an additional layer of protection.

Cloud security:

• Additional encryption (on Azure): We offer the option of double encryption with customer-managed keys for an additional layer of protection.
• AWS (Frankfurt): Provides encryption at rest, protection against DDoS attacks, and continuous monitoring. Compliant with ISO 27001 and SOC 1, 2, and 3.
• Azure (France and Sweden): Applies multi-factor authentication, identity management, and round-the-clock monitoring. It does not store customer data from the OpenAI service and ensures that data remains within EU borders.

Model training and data usage

Praven Intelekt is fully transparent about how your data is used in connection with our AI models.

Data is not used to train OpenAI models: Whether you are a free or paid user, your data is not sent to OpenAI and is not used to train their models. This includes both your prompts (the messages you send to Praven Intelekt) and any documents you upload to our platform. We use the dedicated, secure Azure OpenAI Service, which guarantees that:

• Data is not accessible to other customers or to OpenAI.
• Data is not used to train Azure or OpenAI models.
• Data is not used to improve services.

Azure OpenAI runs entirely within the Microsoft Azure environment and does not interact with OpenAI’s public services, such as ChatGPT or their API. You can learn more about the Azure OpenAI Service measures for data protection, security, and privacy here.

Praven Intelekt does not use any data from customers’ uploaded documents or prompts to train our models. Data is processed confidentially and deleted in accordance with our internal data retention policy.

Pseudonymization and anonymization of data

To safeguard the confidentiality of your personal data, we apply additional protective mechanisms.

We do not use personal data to train AI models: No personal information provided by users is used to train our algorithms.

Data anonymization: Before any data is used for internal purposes, we apply robust anonymization methods to remove all identifying information. We use both neural named-entity recognition (NER) models and rule-based models.
This ensures that personal or sensitive information is never disclosed or used inappropriately.

We regularly update our practices in line with applicable law and industry best practices. For questions or comments about our security measures, please contact us at: [email protected].

For more information about how we process personal data, please see our Privacy Notice.