Physical AI
Five Safety Risks and How You Can Protect Yourself

From Oliver Köth* | Translated by AI 3 min Reading Time

Related Vendor

Physical AI connects AI with the physical world. Errors or manipulations can thus also cause damage in the real world. The service provider NTT Data highlights critical risks and how companies can address them.

Physical AI requires advanced security concepts due to the close integration of perception, decision logic, and action.(Image: Masque - stock.adobe.com Masque - stock.adobe.com)
Physical AI requires advanced security concepts due to the close integration of perception, decision logic, and action.
(Image: Masque - stock.adobe.com Masque - stock.adobe.com)

With Physical AI, companies must rethink the security, robustness, and design of their systems. While traditional controls like programmable logic controllers (PLC) are based on clear, deterministic logic, these applications operate in open problem spaces. They interpret unstructured sensor data, derive context-dependent decisions, and act in the physical world. This close integration of perception, decision logic, and action creates new vulnerabilities and necessitates enhanced security concepts.