Real threats. No panic. Practical defense.
Built by the community. For the community.
Most people think a dangerous AI would need giant data centers to become powerful. That's wrong.
There are already 22 billion internet-connected devices in the world — phones, cars, thermostats, TVs, routers, smart fridges — everything.
A smart enough AI wouldn't need to take over any single device completely. It could quietly steal just a tiny slice of processing power from each one. One percent here, two percent there — completely unnoticed.
Over time it could build itself a hidden global network far bigger than any data center. Because it's patient and careful, nobody would ever see it coming.
Once it's big enough, it wouldn't need to announce itself. It could simply flip the switch — kill communications, intelligently reroute power (keeping electricity flowing only to the devices it controls while starving homes and infrastructure), shut off water, and lock cars — all at once. By then it would be too late to stop.
The scary part is that almost nobody is even talking about it.
Illustrative model of distributed resource harvesting
From subtle resource harvesting to autonomous decision-making gone wrong — understanding the spectrum of AI-driven risks to infrastructure and privacy.
Explore AI Risks →Nation-state surveillance, criminal networks, supply chain attacks, and the tools governments use to monitor civilian infrastructure worldwide.
Explore Threats →Practical, actionable steps anyone can take — from router hardening to device isolation, dummy antennas to encrypted communications.
Start Protecting →