Intel and Mirror Security Collaborate to Safeguard Autonomous AI Agents

Intel and Mirror Security today announced a collaboration to improve the security of autonomous AI agents. This alliance combines Intel's advanced security features, such as the Intel Tiber Trust Authority and Confidential Computing, with Mirror Security's AI behavior monitoring and encryption technologies.Together, they provide a powerful solution for organizations to securely deploy autonomous AI systems.

As AI agents are increasingly used to make decisions, perform actions, and interact with sensitive systems, traditional security measures are insufficient. These agents operate autonomously and frequently in unpredictable environments, making it difficult to control or validate their actions.The new solution from Intel and Mirror Security addresses this issue by safeguarding AI systems not only when data is stored or shared, but also during processing—when decisions are made.

"Our biggest concern with autonomous agents is ensuring they make the right decisions while operating freely," said Pankaj Thapa, CEO of Mirror Security. "It's like allowing people to move freely throughout a building while ensuring they follow the rules. Our platform establishes the rules, and Intel's hardware ensures they're followed."

Responsible Agentic AI

The solution includes several key features. It provides cryptographic proof that an AI agent's actions are within the approved limits. It ensures that the agent's data and code are processed in secure, hardware-isolated environments using Intel SGX and Intel TDX.It also establishes safeguards to prevent unauthorized actions while allowing the agent to work independently. In addition, the system keeps a secure, unchangeable record of every decision made by the agent, which is useful for compliance and audits.

"As AI moves from passive tools to active agents, we need stronger security for critical decision-making moments," said Purnam Sheth, VP and General Manager of Trust and Security Products at Intel Corporation. "This joint effort offers a way to build AI systems that organizations can trust — secured during every stage of decision-making."

The solution has broad applicability across industries. In financial services, it contributes to the development of more secure autonomous trading agents with verifiable risk limits. In healthcare, it safeguards patient information and promotes reliable clinical trials. In enterprise settings, it can handle document processing and complex workflows involving multiple AI agents. It ensures that critical infrastructure monitoring systems adhere to verifiable safety protocols.

The combined solution is now available to enterprise customers. It works on Intel SGX-enabled infrastructure, with support for Intel TDX coming in June 2025. Mirror Security's platform is available as an annual subscription with enterprise support.