Sysdig

EtherRAT marked a shift from opportunistic exploitation to long-term, stealthy access. This implant goes far beyond cryptomining and credential theft — designed to stay hidden, maintain access, and blend into normal activity. The result: a resilient, hard-to-detect implant built for sustained access 📈 — not quick wins. In this session, Crystal Morin and Michael Clark from the Sysdig Threat Research Team discuss how EtherRAT works, what makes its tradecraft unique, and what defenders need to watch for — from blockchain-based C2 resolution to fileless-style execution and aggressive persistence. 🎥 Watch the on-demand webinar to see how it operates and how to detect it in your environment: https://okt.to/B89i1f #CloudSecurity #ThreatResearch

Attacks are moving faster than most teams can respond. As disclosure-to-exploitation windows collapse, supply chains weaken, and AI introduces new blind spots, risk is accelerating fast. On April 9, join Sysdig Threat Research expert Crystal Morin and CISO in Residence Conor Sherman for a live breakdown of what’s actually impacting risk right now: → How fast attackers are operationalizing new vulnerabilities → What recent supply chain attacks reveal about “trusted” tools → Where AI is quietly expanding your attack surface Security dominated the headlines in March. Come get the context behind the news and what to do next. Bring your questions. Leave with answers you can act on.

People are securing AI prompts. 🤖 Attackers are targeting the infrastructure. ⚔️ AI systems run on cloud workloads that concentrate data, credentials, GPUs, APIs, and access — making them a high-value target for attackers. That’s why AI security needs to go beyond prompt injection and application-layer controls. In this article, we break down: 🔹Why AI infrastructure is becoming a primary attack surface 🔹How threats target pipelines, model registries, and inference endpoints 🔹Why posture management, lifecycle security, and runtime protection all matter AI is not just an application - it’s also a workload. And it needs to be secured like one. 👉 Learn more: https://lnkd.in/gfyCsmNV #SecurityforAI #CloudSecurity

☁️ Serverless containers simplify infrastructure management, but they don’t eliminate runtime threats. Workloads running on AWS Fargate can still execute malicious binaries, launch unauthorized processes, or introduce malware after deployment. Sysdig now extends runtime malware detection to AWS Fargate, helping SOC and security teams: 🔹 Detect malicious binaries written to container filesystems 🔹 Identify suspicious processes during runtime execution 🔹 Maintain consistent visibility and detection across serverless and container environments As teams adopt serverless container architectures, runtime security needs to evolve with the cloud. 👉 Read the full article: https://lnkd.in/gAyqqJkh #CloudSecurity #AWS #Fargate #ContainerSecurity

AI coding agents are already running on developer machines and inside CI/CD pipelines. 🤖🔍 Across tools like Claude Code, Gemini, and Codex. And most teams don’t actually see what they’re doing. 👀 What’s happening: ➝ Agents execute shell commands on your machine ➝ They read and write files using your OS-level permissions ➝ They make outbound network calls to APIs and external services ➝ One “response” can trigger dozens of system-level actions ➝ Sensitive data (tokens, configs) often lives in predictable local directories 💥 What this means: ➝ Agents behave more like autonomous users than traditional software ➝ Their behavior is prompt-driven, not fixed or predictable ➝ Built-in guardrails operate within the same trust boundary ➝ Prompt injection can turn normal inputs into malicious actions ➝ Traditional tools weren’t built to track this kind of runtime activity The reality: You don’t see the prompt that caused the action. But you can see what actually happened at runtime. 👉 That’s why kernel-level visibility is critical for detecting risk in agentic AI environments. Read the full breakdown from the Sysdig Threat Research Team: https://lnkd.in/g3b9uKjJ #SecurityforAI #CloudSecurity

Two days into #RSAC, and one thing is clear: speed isn’t just increasing — it’s redefining cloud security. 🚀 Today, attacks unfold in minutes, and defenders are expected to keep up in real time. That’s exactly why runtime visibility and real-time detections matter more than ever, especially as AI-driven workflows execute actions that are harder to predict and monitor. If you’re in San Francisco for RSA, keep the conversation going with us tonight at ‘Cocktails & Cloud Talk with Sysdig & Cloudsmith’. 🍸 Join us for drinks, bites, and a chance to connect with fellow security, platform, and devOps leaders. See you there: https://okt.to/cnMmWY 📍https://okt.to/De4dvJ

We just dropped new Falco runtime detection rules for AI coding agents. 💧💧💧 These agents can write code, run commands, and access sensitive data faster than you can see or control. But do you know what they’re actually doing? 🤖 👉 If you’re at KubeCon + CloudNativeCon Europe, come visit us at booth 671. We can show you how we're detecting and stopping suspicious AI agent behavior in real time. — If you want to go deeper into runtime security, join Iacopo Rozzo (Sysdig) and Aldo Lacuku (Kong) at 12:00 PM CET for their session, “In Falco's Nest: The Evolution of Cloud Native Runtime Security.” You’ll learn: ➝ The latest advancements and strategic direction of Falco ➝ How to secure runtime environments at scale with Falco Operator ➝ The newest and most critical features in Falco See you there 🔗 https://okt.to/rOlQ5t #KubeCon

🚨 The Trivy attack was just the beginning. Within days of the March 19 compromise of Aqua Security’s Trivy GitHub Actions, the same credential-stealing payload appeared in a second, unrelated action: Checkmarx AST. No new exploit — just stolen CI/CD credentials reused to compromise additional GitHub Actions. 👀 What we know: ➝ The same credential stealer was used in both attacks ➝ Credentials were harvested from CI runners and reused to extend the attack ➝ There were identical execution patterns across different tools (Trivy, Checkmarx) ➝ Vendor-specific typosquat domains were used to blend into expected traffic ➝ Consistent exfiltration technique (encrypted archive via curl POST) 💥 Why this matters: ➝ One poisoned action can expose tokens ➝ Those tokens can poison other trusted actions ➝ Each wave looks “different” (new repo, new domain) ➝ Traditional detection (tags, reputation) fails 🛡️ What to do now: ➝ Rotate ALL CI-accessible secrets (including GitHub PATs, cloud service creds, etc.) ➝ Audit GitHub Actions workflows (from Mar 19-23) for references to tpcp.tar.gz, aquasecurity, or checkmarx(.)zone ➝ Search your GitHub organization for repositories named tpcp-docs. ➝ Pin GitHub Actions to full commit SHAs (not tags) ➝ Enable runtime detection on CI runner infrastructure (with Falco or Sysdig Secure) ➝ Monitor outbound network connections from CI runners for curl POST requests ➝ Restrict or disable IMDS access from CI runners 📺 Here’s the reality: From tj-actions/changed-files in 2025 to TeamPCP’s multi-action campaign in 2026, CI/CD supply chain attacks are becoming multi-stage and self-propagating. 👉 Which means: You won’t always catch the compromised dependency. But you can catch what it does at runtime. Read the full breakdown 👇 https://lnkd.in/gVQVmvtZ

🚨 AI has already changed how code gets written. Now it’s changing how it gets attacked. Today, during #RSAC2026, we’re announcing runtime security for AI coding agents to help organizations adopt AI-driven development without sacrificing security. 📣 AI coding assistants are creating new opportunities for technical and non-technical teammates alike; however, with access to sensitive data, credentials, and environments, they also introduce a powerful new attack surface. With real-time visibility into AI agent behavior, Sysdig now detects high-risk actions as they happen, including: 🤖 Installation of new AI coding agents 🤖 Attempts to access sensitive files or credentials 🤖 Unsafe configurations that weaken safeguards 🤖 Malicious activity like reverse shells, binary tampering, or persistence mechanisms As AI agents move from helping developers write code to running critical business workflows, organizations need security that operates at runtime, where risk actually occurs. That’s why we built runtime security specifically for AI coding agents: to enable safe, confident innovation. Learn more 👇 https://okt.to/x2Rk4f

Falco turns 🔟 this year! 🎉 And we’re celebrating the project’s big anniversary with a commitment to its future. At KubeCon + CloudNativeCon Europe 2026 in Amsterdam, Sysdig is proud to announce a $70,000 donation to the Falco project through the Linux Foundation’s crowdfunding initiative. This donation will support the next decade of open source runtime security. Over the years, Falco has grown from an early container visibility project into the open source standard for cloud-native runtime threat detection, with 200M+ downloads and adoption across 60% of Fortune 500 organizations. As Kubernetes becomes the backbone of AI workloads, runtime security has never been more critical. Sysdig’s contribution will help accelerate: ✅ New feature development ✅ Contributor stipends to support maintainers ✅ Technical writing to improve documentation and user experience ✅ New, no-code contribution pathways, enabling contributors of all talents to support Falco If you’re attending KubeCon EU, come see us at Booth 671. We’d love to talk about the future of cloud-native security and celebrate 10 years of Falco with the community! 👉 Learn more: https://okt.to/ojDbwJ #KubeCon #KubeConEU