AI Helped Find a Flaw Putting Up to 70% of Websites at Risk

AI-assisted researchers found a critical flaw in software used by up to 70% of websites. Attackers probed it before the patch. What it means for you.
A security flaw found with help from AI has sent websites around the world scrambling. Some went offline for a while just to be safe.
According to CNN (Oct 21, 2026) and PCWorld (Oct 22, 2026), with The Washington Post and The Wall Street Journal reporting the same figure, researchers using AI tools found a critical flaw in software used by as much as 70% of websites.[1] [2] [3] [4]
What happened
CNN and PCWorld report the flaw could let attackers break into websites and run their own code on the affected servers.[1] [4] Both outlets report it could also let attackers steal data from affected servers.[1] [4]
CNN and PCWorld report the weaknesses had been hidden in the code for about a decade or more.[1] [4] CNN and PCWorld report that turning them into a working attack was extremely complex until AI tools did much of the hard work.[1] [4]
A fix was released on the day it was announced, and website owners were urged to install it immediately.[1] [4] CNN and PCWorld report that some government and company websites in several countries were briefly taken offline while they patched, and that millions of attempts to exploit the flaw were seen within days.[1] [4]
Why it matters
AI helped researchers find the flaw and build a fix. But PCWorld and BleepingComputer report attackers were already probing it four days before the patch was released.[4] [5]
CNN quotes experts warning that not every site will update at once, and that attackers can also use AI to move faster now the flaw is known.[1] The gap between “a fix exists” and “everyone has installed it” is where the danger lives.
For ordinary people, the risk sits with the websites you use: shops, booking sites and portals that hold your details. You cannot patch them yourself, but you can limit what a hacked site could take from you.
How to stay safe
- Website owners: update now. Ask your host or developer if your site uses the affected language and if the Oct 21 update is installed.
- Use unique passwords. If one site leaks, a different password everywhere else keeps the damage small.
- Turn on two-step login. A stolen password alone is then not enough to get into your account.
- Watch your statements. Check your card and bank activity for anything you don’t recognise this month.
The bottom line
AI is now finding serious flaws faster than ever, for good and for bad. The fix only works once it is installed, so update what you run, and protect your accounts as if any site could leak.
I don’t hate AI. I just want it safe.
This article is based only on the published news reports listed below. Dates are as published by each outlet.
FAQ
How many websites are affected?
CNN and PCWorld report the flaw is in software used by as much as 70% of websites. The Washington Post and The Wall Street Journal report the same figure.
How did AI help?
CNN and PCWorld report AI tools helped researchers uncover long-hidden weaknesses and turn them into working attacks so they could be tested and fixed.
Is there a fix?
Yes. CNN and PCWorld report a security update was released and website operators are being urged to install it immediately. PCWorld and BleepingComputer report attackers were probing the flaw before the patch came out.
See the post on YouTube Watch and follow AI Safe Mode on YouTube @AISafeMode and Instagram @aisafemode.
References
- CNN, Oct 21, 2026 (updated Oct 22). Websites shut down globally after AI finds cybersecurity flaw affecting 70% of the internet. Read the report ↗
- The Washington Post, Oct 21, 2026. Websites scramble to fix software flaw that could put much of internet at risk. Read the report ↗
- The Wall Street Journal, Oct 21, 2026. AI just found a bug that puts up to 70% of the world's websites at risk. Read the report ↗
- PCWorld, Oct 22, 2026. AI helped uncover a security nightmare in the web's backbone. Read the report ↗
- BleepingComputer, Oct 22, 2026. Exploitation of the flaw began 4 days before patches. Read the report ↗
