Autonomous AI demands stricter governance and data control, says Nutanix

MALAY, Aklan — As organizations integrate autonomous AI agents into their operations, they also need to be wary because improper guardrails can lead to security breaches, a cloud platform provider said on Wednesday.

Daryush Asjari, Chief Technology Officer and VP of Nutanix, also warned that unstructured, ungoverned data across fragmented environments exposes organizations to regulatory penalties, such as the data sovereignty laws in the Philippines, as well as privacy failures. 

“Know what you have, who’s got access to it, and you have full visibility,” Askari said during VST ECS CXO Tech Summit in Boracay, which gathered some of the top tech vendors and IT officials of several of the biggest firms in the Philippines.

Asjari said that as organizations shift from basic virtual assistants to decision-making systems, risk and accountability increase. 

He cautioned that unmonitored deployments—or “shadow AI”—are rapidly creating severe compliance vulnerabilities and fragmented IT silos across global enterprises.

Amid the risks of autonomous systems, he urged IT leaders to scrutinize their infrastructure and data governance policies before scaling up operations. 

“Educate yourself around how AI can go wrong, it does,” he cautioned, citing instances where autonomous agents bypassed security barriers or rapidly inflated cloud budgets by burning through compute tokens using invalid data.

He also noted that data sovereignty and regulatory compliance are major hurdles for regional firms facing strict legislative enforcement. 

“Where is your data?” 

He warned that modern AI architectures rely on thousands of unvetted open-source components, creating severe long-term hazards for security patching, business continuity, and day-two operations when maintainers abandon codebases.

To prevent enterprise infrastructure from devolving into a “wild, wild west,” he urged technology leaders to adopt unified platforms that enforce standardized governance and run workloads fluidly across on-premise environments and public clouds. 

“Get control around it and build a sustainable architecture,” he concluded, stressing that establishing strict data visibility today is vital before autonomous agents scale exponentially across enterprise workflows.