What happened
Cisco's AI Defense October 23, 2026 release notes introduce MCP Scanning (a documentation page published the same day) and a new 'Tool Exploitation' rule added to AI Runtime guardrails that protects against agent tool misuse; the release also renames all 'validation' functions/screens to 'AI red teaming.'
Why it matters
It's a material expansion of Cisco's AI-security platform: scanning MCP server surface area plus a guardrail rule specifically targeting tool misuse — the attack path where agents and MCP servers are most exposed — from a major network/security vendor.
Applicability
Enterprises running Cisco AI Defense should enable the new Tool Exploitation guardrail and review MCP scanning coverage for their MCP server inventory.