> "As a reminder, there is not a single LLM on the market today that is not vulnerable to prompt injection ... And by and large, companies don't really seem to care."
So far from the truth. I know that there are entire teams that specifically work on prompt injection prevention using various techniques inside companies like Microsoft and Google. Companies do care a lot.
> Based on our research and findings, we assess this to be one of the most technically sophisticated exploits we've ever seen, further demonstrating that the capabilities NSO provides rival those previously thought to be accessible to only a handful of nation states.
> Using over 70,000 segment commands defining logical bit operations, they define a small computer architecture with features such as registers and a full 64-bit adder and comparator which they use to search memory and perform arithmetic operations. It's not as fast as Javascript, but it's fundamentally computationally equivalent.
They create an emulated computer by decompressing an old image format inside a PDF file which has a .gif extension! That is top notch!
Nat was a great CEO. The best that could have happened to Github after the acquisition IMO.
The one time that I won't forget about Github under Nat, was when they stood up for Iranian developers [1]. They went the extra distance to get a permission/license from the US government specially to offer full Github to developers from Iran. Many other companies didn't do something similar.
Technically speaking, this does not prove that an adversarial attack is possible on the CSAM system of apple, Given that apple has another not released neural hash system on their servers which is potentially larger and works better than the one on device.
The more interesting technical question for me is: do collisions transfer across models? or how to find collisions that transfer across models?
I really like the fastai abstractions and their attention to detail. Also their callbacks are amazing. I always look at their implementations for inspiration.
First off, as many know, this is not a single incident. This has happened to many of us (including me) from "sanctioned" countries. (I am from Iran, but live in Germany)
But this makes me think. Many of the services I use are from US companies. Most notable are: Gmail (Google Drive, Google Calendar, etc.), Github, Dropbox, Trello and Toggl. Should I be worried? Should I actually start switching away from these services before they close my accounts and remove all my data without notice?
I have used similar setups in Sublime Text for Python development, the main issue is instability and inefficiency. Installing these set of packages in sublime makes things slower and more unstable.
Multiple times while editing large files my editor would slow down or crash.
One thing which is missing is a "model zoo", a place where people share models (usually well-known) pre-trained, which is very useful for starting to use a framework.