Akamai Report Urges Behavioral Controls for AI Agent Security
A new Akamai Technologies report warns of risks in AI agent deployments, highlighting sensitive data exposure and visibility gaps in software connections.
Akamai Technologies has released a new security report titled 'Speed, Scale, and Non-Human IDs: The Agentic Threat Landscape,' which outlines emerging risks associated with AI agents in enterprise environments. The report recommends that organizations shift their security strategy from merely managing access permissions to actively controlling the behavior of AI agents.
According to the report, enterprises face significant visibility challenges regarding the Model Context Protocol (MCP), which is used to connect AI agents with various software systems. These integration points, if not properly monitored, can create security gaps.
The data highlights the scale of potential exposure within corporate networks. The report found that over 6% of enterprise AI chatbot conversations contain sensitive information. Furthermore, the analysis indicates that more than 40% of enterprise users have installed AI-based browser extensions, expanding the attack surface for organizations.
Quick answers
What is the main recommendation of the Akamai report?
The report recommends that enterprises shift their security focus from managing access permissions to controlling the actual behavior of AI agents.
What risks does the report identify regarding enterprise AI usage?
The report notes that over 6% of enterprise AI chatbot conversations contain sensitive data and identifies visibility issues with the Model Context Protocol (MCP) used to connect agents to software systems.