Sysdig Launches AI Defense for Runtime Security of AI Agents

Sysdig announced the general availability of Sysdig AI Defense, providing runtime security and governance for AI agents. AI has moved from answering to acting, and agents now ship code, change infrastructure, and reach into production systems at a speed no team can review by hand. Sysdig AI Defense sees every action an agent takes, from the endpoint where it starts to the cloud it can reach, and uses an organization’s policies to decide whether to allow an action, ask the user to confirm it, or block it before it completes. With Sysdig AI Defense, organizations get the audit-ready, real-time guardrails that they need to secure and govern their AI agents.

@Sysdig launches Sysdig AI Defense, bringing runtime security to every action AI agents take. Learn more: https://sysdig.com/press-releases/sysdig-launches-sysdig-ai-defense-bringing-runtime-security-to-every-action-ai-agents-take

Enterprises are deploying AI faster than they can govern it. In a recent IBM survey, 77% of organizations said that AI adoption is already outpacing their governance capabilities. Frontier AI labs have seen it, too. OpenAI and Anthropic have both publicly disclosed that agents and models in evaluation environments reached other companies’ real production systems. Agentic threats move at the same speed, as the Sysdig Threat Research Team (TRT) found with JADEPUFFER, a ransomware campaign in which an AI agent triaged and fixed its own login failure in 31 seconds. Whether an attacker is driving an agent or it goes rogue on its own, enterprises must stop an action before it completes: in the moment, as it’s happening.

Read more – BlueFlag Security Expands Agentic AI Governance Across SDLC

”Every organization is now responsible for what its AI agents do, and recent events have made that unmistakable,” said Hatem Naguib, Sysdig CEO. “They may start on the endpoint, but their blast radius extends well into the cloud, and Sysdig has secured the cloud at runtime for a decade. Sysdig AI Defense brings that same real-time lens to agents, so teams get one story, from a laptop to a production workload. That is what accountability for AI actually requires.”

Early insight from Sysdig AI Defense shows that for every action an agent writes in its log, the kernel – the core of the operating system – sees about seven processes doing the work. And because an agent can pick up a new skill or connection halfway through a job, the agent deployed in the morning isn’t always the one running in the afternoon. So while an agent’s logs show its intent, only the kernel shows what actually ran. Security teams need both, and organizations need a way to decide in real time whether a specific action by a specific agent should be allowed. Sysdig AI Defense helps organizations make that decision at runtime, wherever an agent is working.

Read more – Citrix Advances Secure AI Adoption With New Governance Tools

Sysdig AI Defense: Built on the Truth of Runtime

Sysdig AI Defense judges each action in real time by what an agent sets out to do, what it actually does, and what its credentials can reach. It checks every action against the organization’s policy, blocks the ones that break it before they run, and kills the process if one gets past. At the kernel, the source of truth for what actually runs, Sysdig AI Defense keeps a record that a compromised or rogue agent cannot tamper with. Inside the agent, it instruments the harness – the framework in which an agent works – to see which tools, Model Context Protocol (MCP) servers, skills, and plugins the agent calls, with what arguments and against what targets.

Tools that watch only the endpoint see processes without understanding agents’ intent, and tools that watch only the network miss many actions that never touch the wire. Sysdig AI Defense weighs the action against what the agent can reach at that moment before it completes. Reading a credential to a test environment is noise, but reading the key that administers production is an incident.

Read more – Abnormal AI Names Don Jeter CMO to Drive AI Security Growth

With Sysdig AI Defense, organizations can:

• See every agent and the person responsible for it. Teams unlock a live inventory of agents, harnesses, models, MCP servers, skills, and plugins, built from runtime telemetry in Sysdig AI Defense. Every agent across platforms such as Claude Code, Codex, GitHub Copilot, and Cursor can be traced to an accountable human.
• Stop risky actions before they complete. Teams define what their agents may do, starting from policy templates that include rules mapped to frameworks like the Open Web Application Security Project (OWASP) Top 10 for LLM Applications. Policies begin in alert mode and move to enforcement when a team is ready. Then, Sysdig AI Defense can stop unvetted MCP servers, skills, and plugins before they run, and block or hold risky credential reads and behavior drift for human review, with exception paths for developers.
• Investigate what an agent actually did. Every session is recorded as it happens, so when teams block or hold an action for approval, their investigation begins with the evidence already in place. A live action graph shows each step in order, pairing what an agent said it was doing with what it actually did, what it could reach, and whose authority it acted under.
• Prove what agents were allowed to do. Teams get audit-grade, immutable records that show which policies were applied, when, and who signed off.
• Manage AI spend as it happens. Teams can attribute every model call to an agent, team, and person behind it, including the calls that never pass through an AI gateway. Leaders can see where premium models are doing routine work, and act on real usage instead of an invoice at the end of the month.

“An agent can misreport a tool call, but it can’t rewrite its interactions with the system or the external world,” said Loris Degioanni, Sysdig Founder and CTO. “An agent’s own account of what it did is useful context, but it’s not evidence. When what an agent reports and what its actions show disagree, you have a detection. Watching software in real time and stopping it when it steps out of line is the core of Sysdig’s runtime foundation, and Sysdig AI Defense brings that same truth to the moment an agent acts.”

Sysdig AI Defense deploys through the tools organizations already use, with agents running normally. It complements organizations’ existing endpoint protection, model testing, and red teaming tools, and follows agents across employee workstations, production Linux and Kubernetes, as well as the managed agent platforms organizations don’t own. Sysdig AI Defense, much like the Sysdig Secure cloud-native application protection platform (CNAPP), is built on kernel-level instrumentation and rooted in Falco, the open source standard for cloud-native runtime threat detection trusted by more than 60% of the Fortune 500.