Every time I see physical attacks on TEEs, I wish someone would put out a position paper on @hdevalence.bsky.social 's "guy with a glock" model
x.com/hdevalence/s...

I thought this was a cute video too! It sort of skips over the interesting parts of buffer overflows, but it really shines in terms of describing the incident response / mitigation process.

Reminds me of the NSO guy they sent to the audience when Citizen Lab did a talk about Pegasus at USENIX. Really creepy guy, definitely confirmed everything I thought about who would work there.

Glad to see you guys rehosting this ❤️

I'm sure the history goes back further, but in my lifetime, the "Patriot act" sure felt like the start of a congressional race to the bottom.

We're still looking for direct links between code here and gfw behavior but this leak has been a goldmine for censorship research.

Issuing a correction: This is not, as portrayed, likely to be a graph of ChatGPT usage in any significant user capacity. This is from OpenRouter openrouter.ai/provider/ope...

While LLM cheating is a big deal, and I have seen it personally in a widespread capacity, the original claim here is false.

My 5-year plan is for the big 4 + PETS to all at least provide nix templates as an option for artifact evaluation.

On the software side, my current crusade is getting more AECs to use Nix (Inria's tutorial is my goto) nix-tutorial.gitlabpages.inria.fr/nix-tutorial...

There are still more factors (hardware, teaching ML people to seed), but IMO reproducible software is a mostly solved problem.

As someone that did undergrad at a school where the main social hub for computer science majors was the Computer Lab, I really miss it for a lot of reasons. Building out locked-down lab environments also gives a ton of opportunities for fun undergrad projects.

I'm lucky to have most of my TAing be a course on binary exploitation where current LLMs fail horribly unless you know what you're doing. I think a lot about lower div CS, and the only thing I can think of is "just make it absurdly hard and curve" which is pretty much my solution to everything.

Good luck 😢 our one solace is OTF is still around for now.

Pretty much. As you probably know, this particular censorship strategy was deployed during a major Chinese political event to suppress dissent, and I think isn't currently deployed.

"The censors are afraid to incur false positives" is a super key tenet of anticensorship theory. Of course, sometimes it is violated when a government gets a bit too tyrannical (Iran, Russia come to mind)