UPDATE: We were able to replicate the Mythos findings using existing models (GPT5.4) Writeup coming early next week, no BS prompts, it's real reproduction
I will say it again, we used GPT5.4 and Opus, and we were able to autonomously find zero-days in the Linux Kernel (in the last 3 weeks) Mythos is probably better at the task of finding potential issues in code, but imo the threshold for "scary" was reached in December or even
Uh-oh! OpenAI needs to remove ChatGPT 5.4 model for the greater good! It's TOO powerful for consumers! /s I look forward to the write up.
what am I missing? If we know what/where to look for, it should be relatively possible with most of the frontier models, right?
This is like saying, UPDATE: We were able to get out cheese omelet to take like Mythos' cheese omelet. Of course you did, because nothing is fundamentally different about Mythos, it's still the same fucking technology.
Don't miss what's happening
People on X are the first to know.
