Security Overview

Concave AI maintains a multi-layered security framework designed to protect client IP and sensitive data at every stage of the annotation life cycle. Our protocols bridge enterprise-grade digital security with rigorous physical site management. 1. Physical & Personnel Security: Managed Facility: All 75 FTEs operate from a secure, Cloud Factory-style managed facility with 24/7 CCTV monitoring and restricted, badge-only access. Device-Free Workspace: Annotators are required to surrender personal electronic devices (mobile phones, tablets, external drives) before entering production areas to prevent unauthorized data capture. Vetted Workforce: Every team member undergoes background screening, identity verification, and signs a comprehensive, project-specific Non-Disclosure Agreement (NDA). 2. Data Infrastructure & Access Control: Role-Based Access Control (RBAC): We implement the principle of "least privilege." Annotators only see the specific data samples required for their current task, with no ability to view entire datasets. Zero-Download Environment: Our workflow utilises web-based work spaces (e.g., Label Studio, CVAT) where local downloads, screenshots, and "copy-paste" functions are strictly disabled. Encryption: All data is protected using end-to-end encryption—AES-256 for data at rest and TLS 1.2+ for data in transit. 3. Compliance & Auditing: Audit Trails: We maintain granular immutable logs of all platform activities, allowing us to track exactly who accessed which file and when. Anonymization: Whenever possible, we encourage the use of PII-masking or pseudonymization before data enters the annotation pipeline to add an extra layer of privacy. Continuous Monitoring: We perform regular internal security audits and vulnerability scans to ensure our systems evolve alongside emerging threats.