ServiceNow Enhances AI Control Tower with Kill Switches to Combat Agent Sprawl

Severity: Medium (Score: 42.9)

Sources: Theregister

Summary

ServiceNow has expanded its AI Control Tower, evolving it from a governance dashboard to a comprehensive command center for managing AI assets across enterprises. This update addresses the issue of AI agent sprawl, where organizations struggle to manage the numerous AI deployments. The Control Tower now includes capabilities for discovery, observation, governance, security, and measurement, and integrates with 30 new enterprise connectors across major cloud platforms and applications. A demo showcased the system's ability to detect a prompt injection attack on a pricing agent, allowing for automated response through a kill switch to disable the compromised agent. The integration of Veza's access graph technology enhances identity access governance across AI environments. The platform currently maps over 30 billion permissions, ensuring compliance and security across AI systems. This release is part of ServiceNow's Australia platform update, aimed at improving enterprise AI management. Key Points: • ServiceNow's AI Control Tower now features automated kill switches for compromised AI agents. • The system can discover and manage AI assets across multiple platforms, enhancing governance. • Integration with Veza allows for extensive identity access governance across AI environments.

Key Entities

• Prompt Injection (attack_type)
• Australia (country)
• Amazon Web Services (company)
• Anthropic (company)
• Microsoft Azure (company)
• OpenAI (company)
• Claude (tool)
• Google Cloud (tool)
• Traceloop (tool)
• Veza (tool)
• Copilot (platform)