DeepKeep on Tuesday introduced a new AI Agent Scanner designed to help enterprises identify and secure the growing attack surface created by AI agents embedded in business workflows.
The Tel Aviv-based AI security vendor said its latest release provides structured attack surface scanning and discovery for agentic AI environments, where large language model (LLM)-based agents autonomously interact with business systems, data sources, and external tools.
AI agents expand the enterprise attack surface
As enterprises move beyond standalone chatbots toward more autonomous, context-aware AI agents, security risks are compounding.
These agents increasingly integrate with collaboration platforms, financial systems, operational tools, cloud services, and even other AI agents to complete business tasks.
DeepKeep cited industry forecasts projecting that AI agents could make at least 15% of routine business decisions by 2028.
Unlike traditional AI applications with relatively constrained access, agentic systems introduce non-deterministic behaviors and dynamic tool usage, significantly broadening the enterprise attack surface.
“AI agents are no longer operating in isolation; they’re quickly becoming fundamental parts of entire business workflows,” said Yossi Altevet, CTO and co-founder of DeepKeep, in a statement. “But without proper safeguards, their expanding attack surface will rapidly become a massive enterprise liability.”
“At DeepKeep, we are committed to securing agentic AI today and tomorrow, and that means innovating even faster than AI is evolving, starting with our new scanning solution, which offers the immediate visibility and protection businesses need to safely leverage agentic AI ecosystems,” Altevet continued.
This expanded connectivity increases the likelihood of data exposure, tool misuse, unintended actions, and other vulnerabilities that traditional security controls were not designed to mitigate.
Structured scanning aligned to OWASP guidance
DeepKeep’s AI Agent Scanner maps an agent’s threat landscape by identifying connected tools, intended actions, data sources, and potential vulnerabilities across workflows.
The platform generates a visual risk map that aligns findings with the OWASP Top 10 for Agentic Applications, offering security teams a standardized framework for evaluating exposure.
The goal is to help organizations detect and manage risks earlier in the AI lifecycle, during both development and production.
The solution also addresses what DeepKeep describes as a lack of standardized language for defining and securing AI agent structures across vendors and frameworks, which has complicated governance efforts for enterprise security teams.
Runtime protection and framework support
Beyond discovery and mapping, the platform also includes runtime protection for select agentic frameworks.
DeepKeep said it can identify where AI firewalls and guardrails should be deployed based on observed agent behavior, tool access patterns, and data exposure, enabling active risk reduction during execution.
The AI Agent Scanner currently supports several major agentic frameworks, including:
- Microsoft-based frameworks
- Agentforce
- OpenAI Agents
- CrewAI
- Amazon Bedrock AgentCore
- n8n
- Make
DeepKeep said it plans to expand its agentic AI security capabilities throughout 2026, including the release of a dedicated red teaming solution.