Data Privacy & Compliance

As video technology becomes more intelligent and AI driven analytics become increasingly important, privacy, security, and regulatory compliance are no longer optional but essential requirements. Organizations must ensure that sensitive video and metadata remain protected, that access is restricted to authorized users, and that AI is used responsibly and transparently. G‑Insights addresses these challenges by combining advanced video intelligence capabilities with strong privacy safeguards, secure access control, and support for compliance with GDPR and the EU AI Act.

G‑Insights has been developed for use in security-critical environments and combines modern AI technology with high standards of privacy, transparency, and access security.

The solution operates entirely on-premise, ensuring that sensitive video and metadata remain within the client’s infrastructure. At the same time, administrators decide which camera channels are enabled for search and analytics functions and how long data is retained. Retention periods can be defined individually, allowing data to be deleted automatically when it is no longer required.
A role-based authorization concept ensures that users can only access cameras and information for which they have been explicitly authorized. Search results are automatically restricted to the resources assigned to the respective user.

Responsible use is also a core principle of AI-powered similarity search. G‑Insights supports the forensic investigation of persons and vehicles but does not make automated decisions and does not evaluate individuals. The AI models used are dedicated exclusively to semantic search and similarity recognition. Search requests are automatically validated before processing so that requests that conflict with regulatory requirements or attempt to derive sensitive personal attributes can be identified and blocked. Additional security mechanisms, including encrypted communication, authentication, integrity checks of AI components, and the logging of security-relevant actions, support secure and transparent operation.

G‑Insights follows a privacy-focused design approach and operates entirely within the customer’s infrastructure. As a result, video data, metadata, and AI-generated data remain under the customer’s control at all times. The solution does not store complete video streams in its search databases. Instead, it uses only the metadata, best shots, and AI representations required for search and analytics functions.

To enhance security, performance, and traceability, different data types are managed in separate storage areas. Metadata, best-shot images, and AI search information are stored and processed independently from one another. This enables precise control over access rights, retention policies, and storage concepts.

Access to G‑Insights is managed through a central user and authorization system, the Geutebrück Hub. Users are granted access only to the functions, resources, and camera channels for which they have been authorized. Roles, permissions, and resources can be managed centrally and adapted flexibly to organizational requirements.

In addition, every search request is automatically restricted to the camera areas assigned to the user. As a result, users only see results from their authorized areas, both in metadata search and AI-powered similarity search.

G‑Insights enables targeted and controlled data storage. Administrators decide individually which cameras are used for analytics and search functions and which categories of data may be stored.

Retention periods can be adapted flexibly to organizational or regulatory requirements. Once these periods expire, stored information is removed automatically. Furthermore, the system continuously monitors storage utilization and automatically cleans up data when required. This prevents data from being retained longer than necessary.

G‑Insights uses artificial intelligence exclusively to support forensic search and analysis processes. The AI is used for semantic search and similarity recognition of persons and vehicles. Automated decisions, evaluations, or recommendations are not performed.

To ensure the responsible use of AI, search requests are validated before processing. Use cases that conflict with regulatory requirements or attempt to derive sensitive personal attributes are automatically detected and blocked. In this way, G‑Insights supports the secure and compliant use of AI in security-critical environments.

The AI models used in G‑Insights are employed exclusively for analyzing image and text inputs within the context of similarity search. The models do not generate independent content, make decisions, or retrain themselves automatically during operation.

Additional integrity mechanisms ensure that only approved and unmodified model versions are used. This guarantees transparent, reproducible, and controllable behavior of the AI components.