82% Indian Firms Fear AI Agents Will Breach Security by 2027

The report from security and AI operations company Rubrik said the lack of visibility leaves organisations unable to secure identities that are already making decisions, taking actions, and interacting with critical data.

Non-human identities tied to agents are proliferating faster than enterprises can track or govern them, forming what researchers describe as a "shadow workforce."

These identities often operate with persistent access and limited oversight, creating new pathways for misuse, compromise, and lateral movement, the report noted.

Over 82 per cent of Indian respondents reported that agents require more manual oversight than how much they save in efficiency, while 81 per cent lack the ability to roll back agent actions without system disruption.

Nearly nine in ten leaders expressed concern about meeting recovery objectives as agent-driven threats increase.

"The threat itself is accelerating. Nearly half of respondents expect agentic systems to drive the majority of attacks in the coming year, reflecting a broader shift in how adversaries operate," the report said.

Autonomous systems compress timelines, scale attacks, and blur the line between insider risk and external compromise, it added.

Around 66 per cent of Indian respondents believed that AI security advice is still too theoretical or early-stage to be practical.

Further, 38 per cent of Indian organisations expect that up to 50 per cent of cyberattacks in the next 12 months could be driven by agentic AI.

Globally, organisations are operationalising autonomous systems without the controls required to govern them, introducing a gap between innovation and security.

The gap is compounded by identity sprawl, even as the operational promise of AI agents is under strain, the report said based on a survey of over 1,600 global IT and security leaders.

"AI disruption is real and rapidly accelerating in India, yet many organisations lack the visibility, control, and restoration capabilities required to securely manage AI-driven environments," said Ashish Gupta, Managing Director, India & Head of Engineering at Rubrik.

- IANS