Sysdig (@sysdig) header
Sysdig (@sysdig) profile picture

The leader in real-time cloud security

San Francisco, CA
Joined January 2014
Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)Photo by Sysdig (@sysdig)
🚨 Java4Shell (CVE-2026-114938): researchers turned a heap overflow in OpenJDK’s AVX-512 Base64 decoder into pre-auth RCE in products like Keycloak, Jira, and Jenkins. Affects 17, 21, 25, and 27. Patch now. No in-the-wild exploitation confirmed yet. Learn more from Sysdig TRT: okt.to/e0PTdu #CloudSecurity #ThreatResearch #Java
1
3
2
363
“These LLMs can rewrite the rules of the harnesses that are supposed to govern them.” Sysdig CISO Conor Sherman on why AI agent governance has to live outside the harness. Watch the full LinkedIn Live: okt.to/4JvoRx
2
2
5
693
CVE-2026-119618: a critical heap corruption flaw in Apache httpd’s mod_http2 (2.4.55–2.4.70) can be triggered over HTTP/2 with no auth. Apache says it can lead to RCE. Patch to 2.4.71 or disable HTTP/2. okt.to/h8jvmM
1
6
430
The board question on AI is coming. Get your answer before November 17. ⏳ Frank Dickson of Dickson Research and Sysdig’s Blair Howard cover the five questions every security leader should be able to answer, in 30 minutes. 📅 Tue, Nov 17 | 8am PT / 11am ET okt.to/cNCtwV #AIDefense #AIAgents #CISO
1
1
201
Every pickle defense in machine learning has been bypassed the same way. 🥒 Sysdig TRT’s Lydia Graslie analyzed six CVE bypasses across PyTorch, Keras, and picklescan and found the same flaw: the checker and the loader disagreed about what the file was. A pickle isn’t data. It’s a program. okt.to/4vj7AJ #CloudSecurity #ThreatResearch #MLSecurity #RuntimeSecurity
1
1
249
When an agent read a credential file, its own log had no entry 82% of the time. The kernel logged every read. 🔒 One organization, one month of AI agent activity, and what Sysdig AI Defense found. okt.to/GAuFp5 #AIDefense #AIAgents #RuntimeSecurity
1
1
2
319
We talk a lot about speed. The Milan team decided to run it. 🏃 Two teams of four took on The Loop Relay, a 20K relay with skyscrapers, team spirit, and zero latency on the cheering. Shoutout for lacing up. Want to join this team? okt.to/HYlnIT #LifeAtSysdig #TeamSysdig #Milan
1
1
4
247
Detect. Respond. Accessorize. 🧢 Pick up your Lumin cap at booth 168 at KubeCon SLC, and drop into our Falco sessions: 📅 Nov 11, 10:30am | Booth 168: Falco Meets AI Threats (live demo) 📅 Nov 12, 2:40pm | Salt Palace 251 D-F: Falco Never Flies Alone KubeNight Out: okt.to/TSP1K3 #KubeConSLC #Falco
1
2
225
An agent can tell you it did one thing. The real story might be something else. 🔒 Today, Sysdig is introducing AI Defense: runtime security and governance for agents. It finds every agent your team runs, decides what they’re allowed to do, and keeps a record the agent can’t change. okt.to/s9AfEg #AIDefense #AIAgents #RuntimeSecurity
2
3
274
Last chance to register: Thursday’s LinkedIn Live is almost here. ⏳ Can you trust an AI agent’s own account of what it did? Sysdig’s Conor Sherman, Shantanu Gattani, and Ryan Davis take it on live, with a demo. 📅 Thursday, Nov 5 | 12pm ET / 9am PT okt.to/il8UKv #AIDefense #AIAgents #RuntimeSecurity
1
2
2
286
About 50% of AI tools on employee laptops never show up in the agents’ logs. 🔒 Frank Dickson of Dickson Research and Sysdig’s Blair Howard take on the five AI questions every security leader should be able to answer. 📅 Nov 17 | 8am PT / 11am ET okt.to/ewzMrm  #AIDefense #AIAgents #CISO
1
1
6
296
From Raleigh to Milan, the Sysdig team went all out for Halloween. 🎃 Masquerading is usually something we detect. This week, we celebrated it. Zero false positives. Several very high-severity outfits. 👻 Want to join this team? okt.to/zHFr7D #LifeAtSysdig #TeamSysdig #Halloween okt.to/cP4dCZ
1
2
215
Sysdig Headless Cloud Security won the Agentic AI Security Innovation Award in the 2026 CyberSecurity Breakthrough Awards. 🏆 In an industry competing on dashboards, we broke through by removing them. okt.to/BfXzOs  #CybersecurityBreakthrough #HeadlessCloudSecurity #AIAgents #Sysdig
1
2
4
264
🚨 Three bugs. One week. The new patch window is shorter than the news cycle. 🐘 PHP2Shell: unauthenticated RCE in PHP core, exploitation confirmed 🟢 Node2Shell: Node.js Brotli bug, also being exploited 🐧 ZDI-CAN-33367 and five other Linux flaws, one with a public root exploit Vulnerability scanning and runtime detection are best friends before patches land. Blog links in the replies. 👇 #CloudSecurity #ThreatResearch #PHP #NodeJS #Linux
2
2
5
572
Mythos access is expanding. Your vulnerability management program doesn’t need a rewrite. 🔒 More defenders can find vulnerabilities faster, but discovery isn’t exploitation. The fundamentals still hold. Blog: okt.to/oSDBrm Last chance to register for “Mythos, Decoded” this Thursday, Oct 29 at 8am PT / 11am ET: okt.to/5UTR1P #CloudSecurity #VulnerabilityManagement #AIAgents okt.to/ZRPbfw
3
5
396
🚨 Node2Shell (CVE-2026-111497, CVSS 9.2): a public repo for this Node.js Brotli flaw appeared about five hours after the fix shipped. No confirmed in-the-wild exploitation yet, but the patch-to-PoC gap was about a day. Patch in order of reachability: start with services that decode Brotli you don’t control. okt.to/3pTBzk  #CloudSecurity #ThreatResearch #NodeJS #RuntimeSecurity
2
2
5
685
🚨 A working root exploit for a Linux kernel bug is public. No CVE. No merged fix. ZDI-CAN-33367 is a gap in the Dirty Frag fix, with root confirmed on current Ubuntu, Debian, and AlmaLinux. Restrict unprivileged user namespaces and enforce seccomp RuntimeDefault while you wait. okt.to/08psRf #CloudSecurity #ThreatResearch #Linux #RuntimeSecurity
7
34
10
138
21,135
AI agents are writing code, touching production, and acting with real credentials. So who’s watching what they actually do? 🔒 Join Sysdig’s Conor Sherman and Shantanu Gattani on LinkedIn Live to talk runtime security and governance for AI agents. Live demo included. 📅 Nov 5 | 12pm ET / 9am PT okt.to/1h0YSs #AIDefense #AIAgents #RuntimeSecurity
1
3
248
🚨 PHP2Shell: two critical CVEs in PHP core (CVSS v4.0 9.2). Exploitation attempts began four days before the patch was released. An unauthenticated attacker can run code with HTTP requests alone. The vulnerable code dates to PHP 5.4 in 2012, so even clean application code is exposed. Patch first and assume breach. okt.to/xlQvRe #CloudSecurity #ThreatResearch #PHP #RuntimeSecurity
3
4
363