Read @edcumming's piece on Heston, then this reunion. The camaraderie only decades of toil in pursuit of perfection can give. “We all went up to Gettysburg, the summer of '63. Some of us came back. And that's all except the details.” https://t.co/Xk5MLInr5C
Hard to overstate how big of a deal this is.
tl;dr — ANY unpatched Linux server with even a single TCP port exposed to the internet (say for HTTP or SSH) and MPTCP active (default for many Go web servers) is vulnerable to complete takeover.
Great line from the CTO of Proton. I’m paraphrasing and stealing it :
Your job is not to use agents to help you write code. Your job is to lead the agents to help them write code.
An internal model at OpenAI has produced a claimed refutation of ETH! Russell Impagliazzo, Mohan Paturi, and I have been asked to help organize an open and independent examination of this claim. 1/4
‘In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache’
There is an (unpatchable?) supply-chain zero-day in PyTorch autograd that allows a regularization term to use the model's weights as RAM and SGD as its CPU clock to run any computation during training.
@dens All the feels for the coolest social app ever made, @dens and team. Building 4sqwifi way back in 2011 was so fun. Chapeau and thanks for all the fish.
Like Hemingway: first slowly then all at once.
With about-equal-frontier-swappable intelligence edge boils down to discoverability — tools, harness, data.
Few months ago Mythos could only find a single vulnerability in curl repository, now it basically became a CTF.
What has changed? Better models? No.
Better harnessing and steering by subject matter experts.