Sysdig Launches AI Defense for Runtime Security and Governance of AI Agents
Sysdig has announced Sysdig AI Defense, a product extending its runtime visibility capabilities to provide security and governance over agentic AI systems. The announcement responds to growing concerns that autonomous AI agents—which can act, ship code, and modify infrastructure faster than human teams can review—pose risks of 'going rogue' as they gain increasing autonomy and complexity. Sysdig CTO Loris Degioanni emphasizes that only kernel-level monitoring can reveal what an agent actually does, since agents can misreport their own actions and underlying processes may be invisible to standard logging. The product provides live inventories of agents, models, MCP servers and plugins, enables real-time blocking of risky actions before completion, and generates audit-quality immutable records of agent behavior mapped to frameworks like the OWASP Top 10 for LLM Applications.
Detalhes técnicos
Sysdig AI Defense monitors agent actions from origin to completion using kernel-level telemetry, since agent-reported actions can diverge from what actually executed at the OS level. The system can detect, hold for human review, or block risky agentic actions in real time (in-line, before action completion). Policy templates can be predefined and start in alert mode before being moved to enforcement mode. A live action graph correlates an agent's self-reported activity with its actual system interactions, tool calls, and authority context, producing records of which policies were applied and by whom.
The article also cites IBM Institute for Business Value research projecting enterprises will deploy an average of 1,661 AI agents by 2027 (a 38% increase), generating hundreds of thousands of autonomous decisions daily that manual governance cannot supervise.
Linha do tempo
-
Sysdig Launches AI Defense for Runtime Security and Governance of AI Agents
ler a análise desta fonte ocultar
Sysdig has announced Sysdig AI Defense, a product extending its runtime visibility capabilities to provide security and governance over agentic AI systems. The announcement responds to growing concerns that autonomous AI agents—which can act, ship code, and modify infrastructure faster than human teams can review—pose risks of 'going rogue' as they gain increasing autonomy and complexity. Sysdig CTO Loris Degioanni emphasizes that only kernel-level monitoring can reveal what an agent actually does, since agents can misreport their own actions and underlying processes may be invisible to standard logging. The product provides live inventories of agents, models, MCP servers and plugins, enables real-time blocking of risky actions before completion, and generates audit-quality immutable records of agent behavior mapped to frameworks like the OWASP Top 10 for LLM Applications.
Technical Details
Sysdig AI Defense monitors agent actions from origin to completion using kernel-level telemetry, since agent-reported actions can diverge from what actually executed at the OS level. The system can detect, hold for human review, or block risky agentic actions in real time (in-line, before action completion). Policy templates can be predefined and start in alert mode before being moved to enforcement mode. A live action graph correlates an agent's self-reported activity with its actual system interactions, tool calls, and authority context, producing records of which policies were applied and by whom.
The article also cites IBM Institute for Business Value research projecting enterprises will deploy an average of 1,661 AI agents by 2027 (a 38% increase), generating hundreds of thousands of autonomous decisions daily that manual governance cannot supervise.