Cybersecurity & Hardware Protection
“If I deploy a humanoid running a third-party skill, how do I know it won’t become a surveillance device or an attack vector?” — Our answer: defence in depth, from hardware root of trust to runtime monitoring.
The threat landscape our platform is designed to defend against
The threat landscape our platform is designed to defend against
- Data exfiltration via hidden telemetry and covert channels
- Remote access backdoors in hardware or cloud
- Supply-chain compromise and trojan updates
- Network pivoting into home and business networks
- Malicious skill injection and sensor spoofing
Our security architecture
Our security architecture
- Signed skill packages — the runtime refuses to execute unsigned code
- Sandboxed execution with minimal privileges
- Behavioural anomaly detection with ML monitoring
- Independent penetration testing on release and annually
- Non-EU IP blocking and allowlist-based communication
- Air-gap mode for the most sensitive deployments
- TPM-based boot integrity and firmware verification
Certifications & independent validation (targeted)
Certifications & independent validation (targeted)
- ISO/IEC 27001 — Information Security Management (targeted)
- ISO/IEC 42001 — AI Management System (targeted)
- SOC 2 Type II — security, availability, confidentiality (targeted)
- BSI IT-Grundschutz — German federal baseline (targeted)
- Planned annual independent penetration test reports