Hmm.

Now, let's get nerdy about this.

Low-grade urea is what you use for fertilizers, and so you're putting it on your crops.

It's a regularly used fertilizer.

High-grade urea is what was used to create diesel exhaust, exhaust fluid, which is what was used for 80 to 90% of Australia's trucking fleet, for example.

We first heard about this by industry calling us up and saying, "Hey, we're a bit worried about these China export controls on urea.

Um, we've heard from some providers that they're not able to get the stock." And then we saw indications that other countries are starting to like scramble to fly to different countries and take bulk ur- uh, urea and diesel exhaust fluid back to their jurisdictions.

Mm

What do you see as the pros and cons at this point of encouraging greater US open source release?

To start that, I might just briefly touch on the drivers of this big open source discussion in the US context, because I really liked your point earlier, David, of like everyone's taking the open AI incident and saying, see, I'm right for a broad range of perspectives.

But I think the agitation and efforts from the open source community in the US is coming from a few things.

First, the approach that was taken to Fable of like restricting access until, you know, certain conditions were met, which is government playing a heavier-handed role in AI.

And then there's been rumours that the White House might start restricting open source AI or open-weighted AI.

And there hasn't been very much credibility put behind those yet, but they have been reported.

And so I think the open AI incident was like another wave that these people machinations and this movement was getting, yeah, a lot of momentum before that incident.

where do you stand on that i mean is that a worthwhile use of our effort and time and money or do we risk just duplicating things that might remain available say open source for this foreseeable future much more cheaply anyway

And what, what are the risks that we're protecting against? Because as you mentioned, there's some common infrastructure, maybe some differences if we're, you know, trying to prevent cyber risk and Mythos-level things versus preventing AI loss of control.

When we think about what parts of the AI tech stack are most governable or most Observable or controllable or monitorable, compute seems to be the key target here.

And that's because when you think about the other parts of an AI model, so whether it's the data or the algorithms, uh, or the model itself at the end, those things are much harder to control because they're not physical.

They can be copied and pasted.

They can be duplicated.

Um, but compute is like a physical piece of the supply chain.

Uh, it's got a very narrow supply chain.

An initial use case that you can imagine is, let's say, OpenAir and Anthropic want to do a mutual verification regime between them, where they set up a data center, they run tests on those data centers, they train some models, they figure out, "How can we make claims to this other party using the kind of technologies we talked about today?" That then can create sort of a technical basis that allows governments to have more trust in relying on this technology in a more broad-based way and sync up with efforts that are now already underway that we mentioned earlier between the US and China.