AI governance shifts to runtime checks as always-on agents spread

Singulr AI CEO Shiv Agarwal says enterprises must rationalize overlapping AI controls and check agents' actual runtime behavior, not just their intended permissions.

Agarwal said Singulr has extended agent governance to local AI tools on employee endpoints, and some chief information officers are keeping agents and large language model calls within their virtual private clouds.

The discussion came from Dell Technologies' "AI Leadership Symposium," which theCUBE covered as a paid media partner.