In October 2026, coding-capable AI agents removed the skill, money and organisation needed to launch cyberattacks, putting the internet at real risk of being knocked out by its own users.
In October 2026 the internet got the chaos experts had spent months warning about: a flood of AI-powered attacks.
Both sides speed up
AI finds flaws fast and cheap. Hackers use it to break in, defenders use it to patch. Midweek, organisations scrambled to fix a bug that could hand over as much as 70% of the world's websites.
PHP, the language under much of the web, had two critical holes found by AI. The NSA's advice: if you cannot patch it, take it offline.
The barrier is gone
Open-weight models, especially from China, put quality hacking tools in everyone's hands, and their guardrails come off easily.
No nation-state resources, no crime syndicate, not even coding skills are needed. Type 'can you find what scams they've done before, like spoofing calls' into a chat box and you can steal files from a university.
Everything rides on one network
Banks, power grids, pipelines and government systems all hang off the internet. Your life's savings are just a row in a bank's database.
It seemed rock solid, so for convenience everything moved online. This month the only thing between your money and someone else's is each bank's own security, and that is looking shakier.
People are now talking about going offline, building air-gapped networks, and the internet temporarily ceasing to exist.
Why it matters
For thirty years society moved its critical systems online because it was convenient, betting that nobody could attack at scale. AI has called that bet: the real danger is not a downed website but power grids and banks whose security budgets never assumed every amateur is now a hacker.
Curated from high-quality sources, with concise summaries and key takeaways.
