SonicWall SMA1000 got a CVSS 10. It was disclosed Oct 6 and exploited by Oct 8. Finding it took seconds; patching every client is the actual job.
MSP owners: how many of yours are patched right now?
https://t.co/ooweyF64Gb
RMM abuse jumped 277% last year. Attackers don't pick locks when you leave the master key under the mat.
MSP owners: what's the one thing you've done to lock yours down?
https://t.co/yTpZngLNV1
@dailytechonx Huntress found the second rogue ScreenConnect client uninstalling the first to cover its tracks. MSPs: are you allowlisting your own ScreenConnect instance, or just the app name?
@SecAlertsCo Patching is the easy part. The hard part is knowing which of your 30 clients is still running SSM On-Prem. Could you answer that right now?
@directorcia Agreed. I can teach someone the stack in a month, but I can't teach curiosity in a year. What's the best interview question you've found for spotting it?
@hnshah Security is the extreme case. Mid-market teams buy 30 tools and staff zero people who know all 30. The interface was always the attack surface. When software learns the user instead of the other way around, the misconfig era ends.
@andrewchen Underrated part of autonomy: a clean handoff to humans when it hits the edge case. Every autonomous system will meet its Fleet Week. The winners won't be the ones that never need a human, they'll be the ones that know exactly when to call one, and fast.
@edsim@satyanadella@KeycardAI@ianliving Right frame. Insider-risk programs never trusted the analyst, they trusted the clearance and the audit log. Agents need the same: least privilege, scoped per task, expiring by default. Weird twist: the model is the first insider that can be socially engineered by a PDF.
@ivanainai Give an agent a goal, a browser and no rules of engagement and you've built a very fast private with no sergeant. The fix isn't smarter models, it's guardrails outside the model: scoped permissions, approval gates, full logging.
@CSOonline At this point NetScaler patching isn't maintenance, it's a standing watch rotation. Patching the edge is half the job. If it was exposed before the fix, assume someone already used it and go hunt.
@johnmcclure00 'Assume compromise' is the line most people will skim past. Patching closes the door; it doesn't evict whoever walked through it. Anyone running fleets of these needs a sweep for new accounts, web shells and odd sessions, not just a green checkmark.
@BleepinComputer Attackers now know who's been butchering Bohemian Rhapsody at 1 AM. Joke aside: 'fan database' is exactly the system nobody owns, nobody patches, and everybody forgot was still connected. Forgotten assets are where breaches live.
@vxunderground Underrated ATT&CK tip: read it backwards. Pick a technique, then ask 'would I actually see this in my environment today?' Most teams find a great map and a coverage plan full of holes.
@vxunderground Trust is the most exploited vulnerability in healthcare and it never gets a CVE. 'It came from Bob' beats every filter we've built. If your awareness training stops at 'check the sender,' you're guarding the front gate while the enemy walks in wearing our uniform.
1/ September Patch Tuesday: 973 CVEs, an all-time record. Help Net is calling it the "Patch Apocalypse." Here's how to survive it without losing your mind. 🧵
5/ Even the domain system isn't sacred. Attackers hijacked the .gh, .sl and .as registries and got valid certificates for Google and YouTube domains. Watch the public certificate logs for your own domains.