That’s an excellent question. In this age of AI, it’s clear that the major cybersecurity firms are powerless against future attacks by AI agents. Anthropic’s Mythos is proof of this. It’s essential to understand that, in theory, only the software’s architect has in-depth knowledge of its vulnerabilities.
That is why I have chosen to use an authentication system where humans remain in control of the keys,
"CryptoLogin"
https://dev.to/takouzlo/stop-storing-password-hashes-i-built-a-zero-storage-auth-system-for-the-ai-era-3djg
With this system, the master_secret remains under the user’s control, unless they choose to store it in a cloud accessible to AI. I am convinced that a traditional notebook is the key to success.
I keep my precious passcode, my master_secret, safely tucked away on a small piece of paper in my wallet, far from the prying eyes of the cloud. In this way, I am confident that I can fend off future attacks from our virtual colleagues(AIA). It is regrettable to note that ‘MYTHOS’ has already established this as a fact. 💪🤞🫡😎
Top comments (1)
If I may, I would like to be honest and realistic with you. It is clear that AI agents all learn the same things, but it is also true that they convey this information to us in different ways. I would ask you to please be frank with us. We must admit that, in the near future, they will be able to neutralise any defence in a very short space of time. AI agents are becoming increasingly powerful and fast. Most engineers now use AI agents for their modules. It is regrettable that AI agents have not yet reached their full potential, but it seems this will not be long in coming. The philosophy of work is evolving rapidly, and cybersecurity in particular is undergoing a transformation.