Responsible AI
AI should help people, not replace their judgement or put their data at risk.
Ethical by design
- Never trained on your data Your media is analysed for you, not used to train models for anyone else
- Licensed models and data We have appropriate licences for all the models and data we use
- Privacy by design Run it in your environment, and use Incognito mode so we never receive your media or metadata
- Non-biometric options Where biometrics are not needed, we offer alternatives, such as non-biometric player recognition
- Humans decide Our compliance tools flag risk, and a person makes the final call
- Tested for bias We have run more than 100,000 demographic bias tests on our face matching
Designed for the law
Our automated compliance models don't just find general offensive categories. They are tuned to discover content according to the rules and regulations in your industry.
We can also detect AI-generated video, so you can label it as the EU AI Act requires.
The platform is designed to support the EU AI Act and EU Data Act.
FAQ
What if my region does not permit facial recognition?
We have non-biometric models for sports player recognition.
How do I enable facial recognition and biometric features?
Face recognition is off by default. It is switched on for your deployment, by you as the deployer or by us, only after we issue an activation certificate. The certificate records your legal basis, the safeguards in place and any conditions. Contact us to start the process.
What if we have strict AI policies?
We are confident our platform can comply. We would be happy to work with you to configure it to meet your needs.