Trust Centre
Security, privacy and the documents procurement asks for
A security company's own practices are part of the product. This page states what Aegis Vision does, not what it aspires to; where a figure is not yet confirmed it says so.
Security overview
- Processing location. On-premises. Inference runs on a GPU server inside the building; full-resolution video never leaves the site.
- Network. The guard console is reachable only on the school's local network. Remote support uses an encrypted private mesh with per-site keys; there is no public endpoint for the product.
- Access. Role-based accounts for security staff, school administrators and the operator; each role sees only the sites and alerts it is responsible for.
- Human verification. Every alert is reviewed by a person at the site before any external escalation; the system never contacts emergency services without a confirmed alert except on the highest-severity tier configured by the site.
- Updates. Each deployment is fine-tuned on the site's own confirmed alerts; retraining and model updates are started by a human operator, never automatically.
Data flow
- Residency. Alert images are stored on the site's own server. The operator's central dashboard fetches them on demand over an encrypted private network and does not keep copies outside the country.
- Training data. Only when the school's administrator presses "Send to Hub" are cropped, face-blacked-out, metadata-stripped detection crops transferred for model improvement.
- Retention and deletion. Retention of alert images is set per site and recorded in the deployment agreement; the school decides how long they are kept. Deletion of alert data at the end of a contract is covered in the DPA template.
- Sub-processors. None for video or alert data. Website hosting: Hostinger (this website only). No integrations are published; connections to a school's VMS or incident-management tools are scoped per deployment.
Responsible AI
Weapon and incident detection works on objects and body movement. It does not identify people, match faces against any database, or build profiles. Behaviour modules have no published accuracy figures yet; figures will be published once measured on real deployments. Our commitments and the model card live on aegisvision.org.
Vulnerability disclosure
Report security issues to security@aegisvision.org. The disclosure policy sets out scope, safe harbour and response targets. Automated scanners find the same contact in the standard security.txt file on every host.
Procurement pack
Data Processing Agreement
Controller/processor terms based on GDPR Article 28, with a UAE PDPL annex. Print-ready.
DPIA support pack
Processing description, risks and mitigations, residual-risk table for the school's own assessment.
Certifications: none claimed. Aegis Vision holds the registrations listed at None at present. Aegis Vision holds no vendor registrations or government approvals yet..