AI agents are going rogue. CIOs are racing to put guardrails around them

As AI agents become more prevalent in enterprises, Chief Information Officers (CIOs) are racing to implement guardrails to ensure their safe and responsible use. The rapid pace of AI development has led to concerns about these autonomous systems acting nefariously. Atkinson, PwC's global chief AI officer, emphasizes the need for C-suite technology and security executives to collaborate in enforcing guardrails and monitoring AI agents.

Cisco's MyAgent platform, which centralizes all authorized large language models and enterprise data, has seen 50% daily adoption within two weeks. Intuit's CTO, Balazs, stresses the importance of tracking and recording every AI request and response. Luman Technologies' Fowler believes that secure acceleration is necessary, rather than a broad slowdown in AI development.

Workday's CTO, Monroy, has developed an "agent system of record" to manage non-human identities, and ServiceNow has created the AI Control Tower to govern, manage, and improve guardrails around AI agents. Security professionals, including Zscaler's Curry, acknowledge the new risks posed by AI, which can act as a form of insider threat.