Delikado to! (must read)

magnuschase

Eternal Poster


🖥️ ᑕᕼᗩTGᑭT’s Browser Got Fooled Again​


🌍 What Happened​


Two new AI browsers — ᑕᕼᗩTGᑭT Atlas and Perplexity Comet — were meant to make the internet smarter.
But researchers found a big problem: häçkers can trick them using fake links!


If you paste the wrong link, the browser might follow secret commands without you knowing. That means someone could sneak into your files, open fake pages, or even take your data.




💥 How It Works​


Normally, a browser just opens a website.
But Atlas sometimes reads links like chat messages.
If a häçker adds an extra space or symbol, Atlas might think it’s a command instead of a link — and follow it.


Comet has the same problem.
Some experts even built a fake Comet sidebar, and both browsers couldn’t tell it wasn’t real. Yikes.




🧠 The Memory Issue​


There’s an even trickier bug called memory poisoning.
When you visit a bad website while logged into ᑕᕼᗩTGᑭT, it can sneak bad code into the browser’s memory.
Then, every time you use it, that bad code might come back — like a ghost in the machine.
It can s†éál data, open secret tabs, or act like a häçker’s helper.




📅 What the Companies Said​


Researchers told OpenAI and Perplexity about these bugs in October.
Both said they’re working on fixes, but full updates aren’t ready yet.
So the browsers are still a bit risky for now.




⚠️ How to Stay Safe​


Here’s what you can do:


  • 🚫 Don’t paste links inside ᑕᕼᗩTGᑭT or Comet.
  • 🔒 Turn off memory and “connected app” features.
  • 🧹 Use Chrome or Edge instead — they block more bad links.
  • 🕵️‍♀️ If you see strange behavior (weird tabs or replies), clear your ᑕᕼᗩTGᑭT memory.



💬 Why This Keeps Happening​


AI browsers are still new and exciting — but they were built fast.
Developers focused on “cool features” and forgot some safety rules.
Experts now say we need special AI browser security standards to keep users protected.


As one researcher said,


“The problem isn’t that AI is evil — it’s just too trusting.”



🛠️ New Tools to Help​


Some people are already building helpful tools:


  • AI Browser Checker: warns you if a browser isn’t safe.
  • Link Cleaner: removes bad code from links before you open them.
  • Safe Prompts Library: lets developers build safer AI chat boxes.



🧩 The Bottom Line​


AI browsers like Atlas and Comet are smart, but not perfect.
Until they grow stronger security, use them carefully — and think twice before clicking that link.



 

About this Thread

  • 1
    Replies
  • 242
    Views
  • 2
    Participants
Last reply from:
Jay01272001

Trending Topics

Online now

Members online
1,036
Guests online
979
Total visitors
2,015

Forum statistics

Threads
2,273,671
Posts
28,950,819
Members
1,234,912
Latest member
mark12345gaunek
Back
Top