EU AI Act Compliance

AI System Risk Classification

Kaddio currently uses AI in low-risk contexts such as:

• Transcription and dictation
• Drafting letters from medical records
• Medical data summaries

We have classified these under the limited-risk category in accordance with Title III, Chapter 2 of the EU AI Act.

Transparency Obligations

To fulfill our obligations under Article 52 of the EU AI Act:

• Users are informed when they interact with AI systems, either through UI labels or contextual information.
• AI features are designed with explainability in mind, including a clear summary of what the AI does and its limitations.

Human Oversight

In accordance with Article 14, all AI-assisted outputs:

• Are reviewable by a human prior to any critical decision-making
• Include the option to override or ignore AI suggestions

Risk Management & Monitoring

We follow a documented internal process for managing AI risks:

• Periodic performance evaluations of AI systems
• Logging of AI decisions and user interactions
• Fallback mechanisms in case of AI failure or inaccuracy

Documentation & Technical Files

We maintain internal documentation of all AI systems, including:

• Functional descriptions and intended purposes
• Data used for training or fine-tuning (where applicable)
• Records of design, testing, and deployment

Data & Privacy

All AI features are developed in accordance with our GDPR-aligned data handling policies, including:

• Use of pseudonymized or anonymized data where applicable
• Respect for data subjects’ rights
• Integration with our Privacy Policy and DPA structure

Contact

For more information on our AI systems and compliance approach, please contact: [email protected]