Gartner Urges Rapid Implementation of Safeguards Against Agentic AI Risks in Businesses

As autonomous 'Agentic AI' systems rapidly permeate business environments, Gartner warns of significant governance and security challenges that German companies—and global organizations—must urgently address. Danielle Casey, Senior Director Analyst at Gartner, compared current deployments of these AI systems to "driving without guardrails," highlighting the unpredictability and control gaps that expose companies to operational risks.

A recent ServiceNow report reveals that 90% of organizations experienced security incidents related to Agentic AI in the past year. However, less than 40% managed to resolve these issues within 24 hours, underscoring the difficulty in responding promptly. Despite 57% of security teams maintaining comprehensive records of active agents, nearly a third found unregistered agents operating unnoticed. Moreover, only 19% monitor over 90% of their agents in real time, while 66% lack automated countermeasures to suspicious behavior, and 82% reported unregistered agents within their networks.

Experts emphasize the need for multilayered protections beyond technological fixes. Gartner recommends establishing clear governance structures that distinctly separate governance roles from IT management, defining decision-making rights, responsibilities, and accountability for each agentic AI. Every autonomous agent should have clear goals, measurable success criteria, behavioral guidelines, and escalation paths including emergency shutdown processes. A trial period and controlled integration into continuous development cycles with documentation and activity logging are also crucial.

Gartner proposes a phased implementation: immediate establishment of decision-making rights, behavioral rules, and escalation procedures; protective measures and trial phases within 90 days; followed by enhanced oversight linked to business outcomes in the subsequent year.

Security experts such as Nathan Smolenski from Qualys note that AI technologies expose and amplify existing vulnerabilities, while Microsoft's Mike Yeh advocates for granting autonomous agents minimal access to sensitive networks and calls for greater transparency from AI providers to prevent systemic breaches.

Complementing these findings, Roland Busch, CEO of Siemens AG, recently underscored in a televised interview the importance of German companies adapting to complex challenges including technological security risks. As businesses face increasing pressure from investors and shifting economic conditions, the governance of emerging AI systems remains a critical frontier.

With autonomous AI systems growing in deployment, the call for urgent, structured, and multi-dimensional safeguards grows louder, positioning German enterprises and others worldwide at a pivotal point in reinforcing security and governance frameworks against evolving AI risks.