|
Work item:
|
X.sf-pAI
|
|
Subject/title:
|
Security framework for physical AI systems
|
|
Status:
|
Under study
|
|
Approval process:
|
TAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2028-Q4 (Medium priority)
|
|
Liaison:
|
ITU-T SG13, SG20, SG21, ISO/IEC JTC 1/SC 27/WG 4 and WG 5, SC 42, IEC, IETF, ETSI
|
|
Supporting members:
|
Korea (Republic of), Soonchunhyang University, KISA
|
|
Summary:
|
The rapid advancement of Physical AI systems, which integrate artificial intelligence with sensing and actuation in the physical world, is enabling increasingly autonomous operations across multiple sectors. AI is rapidly evolving beyond purely digital environments and becoming deeply integrated into physical systems. Autonomous vehicles, robotic platforms, industrial automation systems, drones, medical devices, and smart infrastructure increasingly rely on AI-driven decision-making and control.
Unlike conventional AI systems operating in digital domains, Physical AI systems directly affect the physical environment, introducing new risks such as safety hazards, physical damage, and cascading failures with potential societal and economic impacts.
Existing standards, including those from NIST and ISO/IEC, address AI, cybersecurity, or cyber-physical systems separately and do not provide an integrated approach linking use cases, risks, and controls for Physical AI systems. The physical AI systems introduce security challenges that are not adequately addressed by traditional AI security standard.
Furthermore, emerging industry architectures highlight the growing complexity of Physical AI ecosystems, reinforcing the need for standardized risk identification and mitigation. By combining AI technologies with cyber-physical systems, Physical AI creates new attack surfaces spanning sensors, actuators, control systems, communication networks, and the physical environment itself.
Therefore, this draft Recommendation aims to establish a security framework for physical AI systems, including identifying threats and risks, and specifying controls and requirements to ensure the safety, security, and trustworthiness of Physical AI systems.
The security requirements and controls proposed for Physical AI systems can provide valuable guidance to system architects, operators, and security organizations in designing, deploying, operating, and securing Physical AI environments.
This draft Recommendation aims to develop security framework including security requirements and controls for physical AI systems. Since the security requirements are considered as recommendatory or mandatory, it is appropriate to develop them in the form of a Recommendation, not in the form of TR. Such requirements can be used to assess the security of physical AI systems and to verify their conformity with the Recommendation.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-06-10 08:58:38
|
|
Last update:
2026-06-10 16:15:59
|
|
