ASTM International, EnterpriseSG flag five standards gaps in embodied AI

These gaps are hindering the mass-scale adoption of embodied AI.

ASTM International and Enterprise Singapore have identified five standardisation gaps hindering the mass-scale adoption of embodied artificial intelligence (AI) in a white paper released through the Singapore Standards Council (SSC).

The gaps are data governance and privacy compliance, learning from AI safety and reliability in enterprise systems, workplace safety and risk management, interoperability and middleware standards, and quality certification and procurement standards.

Embodied AI refers to the integration of AI into physical systems such as robots, autonomous vehicles, and intelligent machines, allowing them to operate in dynamic physical environments.

The technology can automate complex real-world tasks, raise productivity, and enable new services, which can help to address labour shortages and ageing populations.

Despite the rapid advances, mass adoption remains constrained by the absence of standards on how embodied AI systems are tested, integrated, procured, governed, and deployed in real-world scenarios, the organisations said in a press release.

Titled “A Pathway for Future Embodied AI Standards Development," the paper was produced with A*STAR and the National Robotics Programme.

It was developed after a workshop held by EnterpriseSG in July 2025 that drew more than 140 industry representatives.

ASTM International and EnterpriseSG will take the gaps to their respective technical committees and global member network, spanning robotics, autonomous systems, and AI, to help prioritise future standards development.

Organisations interested in contributing are welcome to contact either party to explore participation.