Cisco Foundation AI, a research and engineering group at Cisco Systems, has unveiled new agentic security tools to protect autonomous AI systems, SiliconANGLE reported on Thursday.

Cisco said the tools reflect a trend in which AI systems are becoming more autonomous and interconnected and are being used across business operations. It said traditional security approaches are difficult to apply as AI-based environments span cloud, internal systems and external data sources.

To that end, it introduced the “Foundation-sec-8B-Inference” model. Unlike AI language models, this model provides multi-step analysis functions specialized for cybersecurity. It provides reasoning processes for threat modeling, attack path analysis and incident investigation, helping security analysts trust and verify conclusions.

Cisco also introduced an “adaptive AI search framework” that goes beyond static query-based search. It allows an AI model to adjust its search strategy each time it obtains new information, enabling a flexible approach to threat intelligence analysis and incident response.

Cisco also released an open-source “Peak Threat Hunting Assistant.” It uses a team of collaborating AI agents to analyze threat actors and techniques and automatically generate customized threat-hunting plans. It is designed so humans can intervene in decision-making, allowing security teams to fully control model and data access, Cisco said.