I break down how software, AI, and systems actually work. Real-world failures, architecture, and security insights.
Space & Sounds
Joined September 2022
- Tweets730
- Following180
- Followers156
- Likes2.4K
Pinned Tweet
You made it here...
I break down how software, AI, and systems really work, and where they break.
Privacy. Security. Real stories behind the code.
If you like understanding what’s actually going on,
you’ll fit in.
The wild part about the GMO Internet ransomware situation isn't just the fifty million dollar demand. It is that the target is actively pushing Tokyo to invoke a brand new statute allowing state actors to hack back and neutralize infrastructure.
This completely blurs the line between traditional enterprise defense and active state-sponsored cyber warfare. If governments start using law enforcement and military assets to hack back against commercial ransomware gangs on demand, the rules of engagement change entirely.
Microsoft is officially pulling the plug on Windows Server 2016 support this coming January.
For anyone running legacy infrastructure, this means zero security updates, no patches for zero-days, and compliance nightmares for anyone stuck in regulated environments.
The API surface changes, dependency rot, and legacy .NET apps usually break when you try to push them to Server 2022 or 2025.
If your systems are still sitting on 2016, the next few months are going to be stressful.
Microsoft Teams is adding AI-powered, context-aware phishing detection.
Crucial move for SecOps as attackers shift from email to chat for credential theft. #infosec
bleepingcomputer.com/news/mi…
Professeur | AI, Systems & Security retweeted
Anthropic added invisible watermarks to Claude text.
Someone built a remover the same day.
The watermark was supposed to survive copy-paste and light edits.
The remover kills it entirely.
Anthropic complied with EU regulations.
Someone complied with market forces.
The watermark lasted eight hours before someone shipped a tool that defeats it.
The detection mechanism Anthropic promised hasn't shipped yet.
The removal tool already has.
This is how regulation works in 2026.
Government mandates invisible mark. Company adds invisible mark. Someone builds tool to remove invisible mark.
All before lunch.
Claude can now watermark every sentence it writes.
Invisible. Travels with copy/paste. Survives light edits.
Introducing Claude Watermark Remover.
Paste Claude text. Kill the statistical mark. Keep the meaning.
Free. No signup. Built today after reading this @paulg post, shipping now.
claudewatermarkremover.app
Most people think turning on “privacy features” means they’re invisible online.
Apple’s Private Relay is a good example.
It sounds like a VPN… but it’s not.
Here’s what it actually does (and what it doesn’t):
Why this matters:
We’re moving from “block everything” to “reduce exposure.”
Private Relay is part of a bigger shift:
Privacy by architecture, not just settings