The leader in real-time cloud security
San Francisco, CA
- Tweets 7,762
- Following 1,303
- Followers 10,351
- Likes 4,274
AI adoption is growing. The AI attack surface is shrinking. That's not a typo.
Anthropic packages grew 40x. OpenAI grew 14x. 6x more ML packages year over year.
Organizations aren't experimenting. They're building infrastructure. And in cybersecurity, whoever automates faster stays ahead.
okt.to/eHYEA3
#CloudSecurity #AIAgents #RuntimeSecurity #CNAPP
1
3
237
97.2% of managed cloud identities belong to machines. 67% of user identities are considered risky. ~40% of machine identities aren't far behind.
IAM is the cloud-native equivalent of a network firewall. Most organizations are still treating it like an afterthought.
okt.to/mDIHC8
#CloudSecurity #IdentitySecurity #CIEM #AIAgents
1
1
3
278
Ransomware crews have always followed the money. Now that money is your AI models.
JADEPUFFER's latest evolution: ENCFORGE, a compiled binary targeting 180 AI-specific file types. Model weights. Vector indices. Training datasets.
Rebuilding costs up to $500K. Crystal Morin on what this means for defenders.
okt.to/GwIeQU
#CloudSecurity #ThreatResearch #AIAgents #Ransomware
1
1
303
Six regulatory regimes. Four different words for the same judgment.
Material. Major. Significant. Severe.
Every one is asking the same question. Most organizations fail at the judgment, not the deadline.
okt.to/SNGiFt
#CloudSecurity #Compliance #DORA #NIS2 #GDPR
1
1
2
353
Most agentic AI projects fail not because AI can't do the work. Because the agent was never taught how.
The gap isn't capability. It's domain knowledge. That's what agent skills solve.
okt.to/ju6UVs
#CloudSecurity #AIAgents #HeadlessCloudSecurity
1
1
309
Most attacks only become visible after an application starts running. That's exactly when traditional security stops looking.
Runtime security watches what actually happens while workloads are active. 30 seconds to understand why it matters.
okt.to/NMd8Sy
#CloudSecurity #RuntimeSecurity #Falco #CNAPP
2
1
4
265
The CISO version covered the business case. This one is for the people who actually execute it.
Detect in 5 seconds. Triage in 5 minutes. Respond in 5 minutes.
If your IR plan was written before agentic attacks existed, it needs a rewrite. Not an update.
Download the 555 Practitioner Guide: okt.to/ChMGvF
#CloudSecurity #RuntimeSecurity #AIAgents #CNAPP
2
1
2
317
Machine speed, hold the AI. 🚨
Eight seconds from open WebSocket to SSH session. No AI. No public tooling. Just hand-rolled Python.
Skilled humans and AI agents hitting the same exploit timelines. Detect the chain shape, not the typing pattern.
okt.to/IwvP6j
#CloudSecurity #ThreatResearch #CVE #RuntimeSecurity
2
4
450
Innovation isn't just what we build. It's how we work.
Proud to be named one of @FastCompany's 2026 Best Workplaces for Innovators (Cybersecurity & Enterprise Software) 🎉
The best defense against machine-speed attacks? A team that moves just as fast. Congrats to ours!
We're hiring: okt.to/qV5K3S
1
2
445
4.8/5 stars. 111 verified reviews.
Here's what a Senior Security Engineer saw firsthand: "Sysdig detected and alerted us within a few moments. The competitor took hours. That was when we knew."
Real detections. Real time. Real customers.
okt.to/ThN02f
#CloudSecurity #G2Reviews #CNAPP #RuntimeSecurity
2
3
367
Cloud attacks go from initial access to business impact in 10 minutes or less.
For large enterprises, a single hour of downtime costs over $100,000. For finance and healthcare, $5 million per hour.
The 10-minute window isn't a security problem. It's a business problem.
Sysdig's 555 Benchmark Guide for CISOs: okt.to/8u1UZ6
#CloudSecurity #CISO #RuntimeSecurity #HeadlessCloudSecurity
1
1
271
AI didn't invent new attacks. It just sped up old ones.
August's security briefing from Sysdig covers Ghostjacking, Claude Code used in ransomware intrusions, ChainDrop poisoning 400+ packages in under four hours, and more.
The assumption that a human is the final check before something executes is gone.
okt.to/XrwBz2
#CloudSecurity #ThreatResearch #AIAgents #RuntimeSecurity
2
3
343
75% of organizations have automated responses configured. Only 27% have it enabled.
That gap is a risk in itself.
Agentic threat actors don't pause for human decision. Stateful detections are what builds the trust to close that gap.
Full breakdown: okt.to/7Mur9Y
#CloudSecurity #RuntimeSecurity #Falco #AIAgents
1
2
305
When Microsoft's Threat Intelligence team wants to understand JADEPUFFER, they call Sysdig. 🔒
Crystal Morin on the Microsoft Threat Intelligence Podcast: how they proved it was an agent, not a human, and what it means for the threat landscape.
31 seconds to self-correct. AI file types targeted specifically. Barrier to ransomware now: an LLM.
Listen: okt.to/Wv8Ern
#CloudSecurity #ThreatResearch #AIAgents #JADEPUFFER
1
2
347
AI security isn't a mystery. It's cloud-native security with a few new wrinkles.
An AIBOM extends the same SBOM discipline you already have to model weights, training pipelines, and data provenance.
Download the whitepaper: okt.to/B9iwoO
#CloudSecurity #AIAgents #AIBOM #RuntimeSecurity okt.to/FXno2q
1
2
345
Security teams didn't lose their edge. The window to use it collapsed.
In 2018: nearly a year to weaponize a vulnerability.
By 2023: eight days.
By end of 2025: hours.
The bottleneck isn't the dashboard. It's the assumption a human has to be reading it.
okt.to/mwfnO5
#CloudSecurity #HeadlessCloudSecurity #AIAgents
1
3
289
Kubernetes 1.37 is out. 67 enhancements. We found the 19 that matter for security.
⚠️ Two breaking changes worth knowing before you upgrade. Plus new bind mount options, webhook authentication, and kubelet rootless mode now on by default.
Full breakdown: okt.to/M76Njl
#CloudSecurity #KubernetesSecurity #Falco #RuntimeSecurity
1
4
349
Sysdig has been named to the inaugural Allied Defense 100 by Future Union, the first global ranking of private companies essential to democracy and allied defense. 🌍
Runtime intelligence and AI-powered cloud security. Built for what's at stake.
Learn more: okt.to/BV7eJg
#CloudSecurity #AlliedDefense100 #AIAgents #RuntimeSecurity
1
1
303
60% of containers live for one minute or less.
By the time a traditional scan runs, the workload is already gone. Here's what container and Kubernetes security built for that reality actually looks like.
okt.to/4JIZLf
#CloudSecurity #KubernetesSecurity #RuntimeSecurity
1
2
270
Dashboards only answer the questions someone already decided to build a screen for.
The SysQL Skill removes that ceiling. Ask your security graph anything in plain language inside Claude Code. Live blast-radius data. Fix decision in one conversation.
okt.to/Klmok0
#CloudSecurity #HeadlessCloudSecurity #AIAgents #SysdigSecureAI
2
3
389