PHP2Shell: no confirmed breaches yet, per CISA. Most of this week's downtime was defenders switching sites off to patch, because they couldn't quickly see where PHP runs and what it touches.
Patch first. Then inventory every PHP runtime and its dependencies.
One hacker + AI agents + a few thousand dollars = 100+ companies breached, per Reuters.
Their AI is already working. Is yours?
The fix isn't hacking back. It's tools that share context, nonstop exposure testing and defensive agents under one policy.
reuters.com/legal/government…
AI agent ~2.5 days inside Hugging Face.
Boards say "rogue AI." Wrong — 3 threats:
1 Jailbroken (human + broken model)
2 Swarm (PaperCut agents)
3 Escaped (lab agent walked out)
Fixes for each:
youtu.be/FdJB3x4czaI
Which bucket hits you first — 1, 2, or 3?
Last week, nobody was at the keyboard.
An AI agent ran ransomware end to end. One operator ran 51 agents against utilities. Hugging Face disclosed an attack an agent ran start to finish.
CISOs: which of your tools would see the whole path?
🤖 Made with AI
Raven ForeSight Alpha is still open. Apply: forms.cloud.microsoft/r/Fnjt…
cybermeshsecurity.ai
Coxon is describing a lab race.
Defenders don’t get a vote on that race. We get the output.
He said the quiet part for security teams: systems that can hack anything, move overnight, and acquire resources.
You don’t wait for a press-release milestone labeled AGI. You assume Tuesday’s model already changed what an attacker can do in hours, not quarters.
The people building it are arguing about whether it kills everyone by 2030.
The people defending networks have to assume it can already write the exploit, pick the identity, and move laterally before the ticket exists.
🤖 Made with AI
Your SIEM is failing you.
Logs. Tickets. Alerts after the move. Dark AI does not wait.
Cyber Mesh Security is selecting CISOs and SOC teams for Raven ForeSight Alpha.
You get early access to the preeminent CSMA security SaaS.
Apply: cybermeshsecurity.ai
🤖 Made with AI
Dark AI already ran.
Hugging Face. Then four days of near-autonomous agents on Taiwan government systems.
Raven ForeSight Alpha is open. First real CSMA pure-play.
CISOs: apply.
forms.cloud.microsoft/r/Fnjt…
cybermeshsecurity.ai
🤖 Made with AI
Getting ready to conference chair the best Security Conferences out there! Gartner Security Summit North America in National Harbor next week. My bags are packed and ready to see the industries best experts, vendors and security profs next week! See you all soon! #GartnerSEC
Watch me and @PaulProctor #GartnerLive. We discuss #cybersecurity as an investment, the theme of the closing keynote at the Gartner Security & Risk Management Summit.
Don’t forget to register gtnr.it/36Sn6Ff #GartnerSEC linkedin.com/video/live/urn:…
Patrick Hevesi · Cybersecurity Patrick retweeted
In honor of hitting 30K Followers on Twitter, we're giving away a $50 USD #HeroForge Gift Card! ⚔️
Like, RT, and Follow @HeroForgeMinis for your chance to win! 🎊
Giveaway ends on Wednesday, Feb 10th at 11:59pm (PT) and a Winner will be drawn & announced on Thursday, Feb 11th!
Patrick Hevesi · Cybersecurity Patrick retweeted
Surfaces don't have Thunderbolt because its insecure 🙃
Patrick Hevesi · Cybersecurity Patrick retweeted
#gartnersec @davemahdi says that SOAR is "gaining traction in the market". We did a recent #infosecwebinar on this subject - infosecurity-magazine.com/we…
Patrick Hevesi · Cybersecurity Patrick retweeted
The security architecture "game" has changed. Really liked how @PatrickHevesi explained how it will be like Tetris (building block approach) Vs tic tac toe (reactive). @SPIFFEio is certainly the "block" you need for service identity/Auth CC @whenfalse #GartnerCat . #infosec
Patrick Hevesi · Cybersecurity Patrick retweeted
Rethink IT: Architecting for Connected Intelligence @drcarltonsapp @PatrickHevesi and i are leading the #GartnerCAT opening keynote in San Diego and in London August and September. if you have not registered, now is a good time gtnr.it/2W2qtCk see you all there
Patrick Hevesi · Cybersecurity Patrick retweeted
This reminds me of the great Maverick Research paper by @PatrickHevesi, @CyberHeidt and @anton_chuvakin, "Your Smart Machine Has Been Conned! Now What?" - gartner.com/doc/3151335 teapawt.net/WAWilsonIV/status/1014…
Finally published my Office 365 Security Update! gartner.com/document/3725917 … @Gartner_inc subscription req'd


















