Responsible AI

1. Your data is used for your project only

We never use client data to train models for anyone else. Data stays in the location you choose, and we delete or return it at the end of the project unless you ask us to keep it for support.

2. Privacy by design

We collect the least data that does the job. Where we can, we process video on the device, blur or drop faces, and store events instead of footage. Before any project that uses personal data, we help you check the legal basis and, where needed, prepare a data protection impact assessment.

3. Humans stay in charge

Our systems support people. They do not replace their judgement on decisions that matter. Alerts can be reviewed, results can be corrected, and every automatic action can be traced back to the data that caused it.

4. Honest numbers

We report accuracy on realistic test data, including where the system is weak. We test across lighting, camera angles and, where people are involved, across different groups, and we tell you about any gaps we find.

5. EU AI Act

At the start of each project we check where the system falls under the EU AI Act: prohibited, high risk, limited risk with transparency duties, or minimal risk. We then build in what that level needs, such as logging, human oversight, documentation and clear notices to users.

6. Work we turn down

• Social scoring of people
• Emotion recognition in workplaces or schools
• Building face databases by scraping images from the internet or CCTV
• Remote biometric identification of the public for law enforcement
• Systems designed to manipulate or exploit vulnerable people
• Weapons or surveillance of individuals without a clear legal basis

7. Sustainable AI

We choose the smallest model that meets the goal, run it on right sized hardware, and prefer edge processing where it avoids sending large amounts of video to the cloud. This lowers cost and energy use.

Questions

Write to hello@zeynoc.com. We are happy to walk through how these rules apply to your project.