The EU Chat Control proposal would be a catastrophe for Europe. It would seal Europeans behind a new digital Berlin wall, cut off from Signal and other e2e messengers. It's embarrassing and dangerous that it's gotten this close already. Time to bail to back to sanity.

"We need to break with the completely erroneous perception that it is everyone's civil liberty to communicate via encrypted messaging services" - @PHummelgaard, 🇩🇰 Minister of Justice Remember: Under chat control, EU politicians would be exempt from being spied on, unlike you.

Something is rotten in Denmark. https://t.co/hN9xIqH1Rh

Keep your friends close and your fish closer

Stop wasting time on CTF challenges. Learn Docker security, EDR evasion, network segmentation, SAML/OAuth flows, WAF configuration, and how to debug production incidents. You'll be 10x more hireable than someone who rooted 500 vulnerable VMs.

Hacker chique LoRa pager, of course sold out already. But makes me want to make my own LoRa mesh chat device using meshtastic or a similar stack. https://t.co/VcvopIh8id

https://t.co/Taanjrz8jT

31337

Many hackers are targeting security researchers by requesting quotes or offering jobs. Here are tips to stay safe, plus red flags to watch for. I'll use a recent submission form we received as an example. 🧵 (1/11)

PerfektBlue Bluetooth Vulnerabilities Expose Millions of Vehicles to Remote Code Execution

🚨 Hackers can hijack your car over Bluetooth. New “PerfektBlue” flaws let attackers run remote code on Mercedes, VW & Skoda—just by pairing. They can track location, record audio, and even reach engine controls. Full story → https://t.co/JBvlpjqCc7

Would not trust a cyber security professional who has never worked as a sysadmin

AI voices need to identify themselves as such, so the other party knows they are talking to a robot. If we know we are talking to a robot, we absolutely do not need AI voices doing emotional vocalisations, or even worse: vocal fry or uptalk. Instant tab close & laptop slam.

Privilege escalation... in chroot. One of those "the firetruck is on fire" situations.

🤓 If you want to learn more about MCP attacks or vulnerabilities, check out this project called Damn Vulnerable MCP. It allows you to experiment on MCP server security through 10 challenges, from basic prompt injection to multi-vector attacks. https://t.co/ny0jOAPJFk

Reverse engineering Wipeout 2097 PSX models + animation = best 90s screensaver ever

https://t.co/pU4f40N2Ci

This is a good example of how *any* untrusted tokens in the input to an LLM, including from the output of a tool (e.g. MCP server call) can hijack the rest of the output.

@GoogleAI But luckily it does come with integration between Gemini & Google Drive. Which is great! Bc Google Drive doesn't even allow you to sort on file size + doesn't tell you the size of folders. AI to the rescue! NOT.

For those of you considering paying for @GoogleAI Gemini Pro to try out Veo2: you get 4 videos per day. By the time you figure out how prompts work you are done for today i.e. don't bother and find another service.