AI transparency
What our systems do, what they deliberately do not do, and how we help you meet your disclosure obligations.
Six commitments that constrain the product
These are engineering defaults and contractual terms, not aspirations.
Video stays on site
Inference runs on the edge device at your location. Raw video is not transmitted to Braingine or to any cloud service — only event metadata leaves, over encrypted channels.
Biometrics off by default
Biometric identification, age and sex estimation, and cross-camera re-identification are disabled on delivery. They are enabled only at your written request, against a recorded legal basis and a completed impact assessment.
No emotion inference
We do not infer emotions, mood, stress, attention or attitude. The capability has been withdrawn from our catalogue and we will not supply it — in any environment, under any authorisation.
The system informs, it does not decide
Every alert is reviewed by a person before action is taken. No Braingine system makes automated decisions producing legal or similarly significant effects on an individual.
Every model is classified and dated
All 497 catalogue entries carry a recorded risk tier, the regulatory reference that applies, the transparency duty owed, and the date they were last screened.
Notices we supply, you post
Where a capability triggers a disclosure duty, we tell you in writing which capabilities are active and provide the site signage and notice templates in English, Spanish, German and French.
Transparency under Article 50
Article 50 of the EU AI Act has applied since 2 August 2026. It requires that people are told when they are interacting with an AI system, and when they are exposed to emotion recognition or biometric categorisation.
The duty to inform sits with the deployer — the organisation operating the cameras. As the provider, our job is to make that possible: we notify you in writing of every capability active in your deployment that engages Article 50, and we supply the site signage and staff notice templates you need to post, in four languages.
How we classify models
Every entry in our model catalogue carries a recorded risk classification, screened against the prohibited practices in Article 5 and the high-risk categories in Annex III. Classification records the worst-case lawful deployment; the binding context for your site is fixed in the impact assessment we complete with you before any biometric capability is enabled.
Where a model is withdrawn, its catalogue entry is retained and marked withdrawn, with the date and the reason — so the withdrawal itself stays auditable.
What we ask of customers
Our agreements require that the platform is not used for any practice prohibited by Article 5, and in particular that no capability inferring emotions of people in a workplace or educational setting is requested or enabled. Customers deploying in publicly accessible spaces must tell us beforehand, and must not operate real-time remote biometric identification for law-enforcement purposes without the authorisation the law requires.
Where we are
Braingine is a member of the EU AI Pact. Our systems are currently deployed in the Americas. We are building the provider documentation required for high-risk AI ahead of the 2 December 2027 date on which the Annex III rules begin to apply, so that it is designed into the product rather than retrofitted to it.
Questions
For AI ethics, transparency or model governance enquiries, write to [email protected]. For personal data and privacy requests, see our Privacy Notice or write to [email protected].