02 — Capability · BC-760.50.50
AI Runtime Protection
Stop attacks on live AI applications and agents: prompt injection, data theft and manipulated tool calls. Output quality belongs to Output Evaluation & Guardrails.
- AI Application Firewall
- LLM Runtime Security
Realized by · 4
Used in · 0
- Not yet placed on a value stream.
Build it · 1
Governed by · 1
- framework NIST Cybersecurity Framework via BC-760
Guided by · 2
- security Never Trust the Network via BC-760.50
- security Security Is Designed In, Then Verified via BC-760.50
- Top · Decomposes into
- No children: this is a leaf capability.
- Middle · Realized by
- P1OneTrust · Guardrail Enforcement and Guardian Agents
- P2Cisco AI Defense · AI Runtime Protection
- P3Cisco AI Defense · Agent Runtime Guardrails
- P4Glean · Glean Protect+
- Base · Build it
- B1Technology · AI Security & Guardrails
Value streams, frameworks, principles, data and events for this capability are in the Plan view.