- 92% of Indian firms faced AI agent-related security incidents in the past year
- Only 36% of organisations contained AI security incidents within 24 hours
- Just 19% assigned a single owner for AI agent security governance
Indian companies are rapidly deploying artificial intelligence agents, but their ability to secure these systems is lagging behind adoption.
As many as 92% of Indian organisations surveyed experienced an AI agent-related security incident in the past 12 months, yet only 36% managed to contain such incidents within 24 hours, according to a new ServiceNow report.
The findings, based on a survey of 500 security leaders, highlight gaps in enterprise preparedness as AI agents increasingly gain access to business applications, data and operational systems.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
The report, titled Deployed but Ungoverned: The State of Agentic Security in India, found that organisations are struggling to establish clear accountability and consistent oversight.
Despite half of the respondents describing their agentic security arrangements as established, only 19% had designated a single individual responsible for AI agent security governance. Just 18% had a unified interface providing visibility across their deployed agents.
Security controls also remain uneven. Only 30% of organisations use automated deployment checks that prevent AI agents from going live before completing security assessments. Meanwhile, 47% can produce a complete evidence package on demand.
Also Read | AI Could Take 20% Of Global Work, But Create Tens Of Millions Of Jobs In India: IBM CEO | Exclusive
The ability to identify suspicious activity has not necessarily translated into effective responses. Although 51% of organisations have automated real-time detection for unusual AI agent behaviour, 67% cannot automatically intervene when an agent acts abnormally.
Another concern is the emergence of unauthorised AI deployments. Around 37% of surveyed organisations discovered agents operating in production during the past year that had never been registered with or disclosed to their security teams.
Kulmeet Bawa, ServiceNow's GVP and Managing Director for India and SAARC, said enterprises must ensure every AI agent is registered, assigned an owner and monitored through centralised systems.
Mr. Kulmeet emphasised that automated controls and continuous oversight would be critical to maintaining trust and cyber resilience.
Also Read | OpenAI Reveals How It Plans To Stop AI Agents From Taking Unauthorised Actions
The findings underline a growing challenge for Indian businesses and that challenge is to deploying AI agents at scale while ensuring their actions remain visible, controlled and accountable.
(With PTI Inputs)
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.