gpm

I don't have such a stated policy... I do have that he both

1. Signed an open letter https://terrytao.wordpress.com/2026/09/11/a-severe-misalignm...

2. Posted a guest post about why they didn't sign said open letter https://terrytao.wordpress.com/2026/09/17/why-i-didnt-sign-t...

If he thought it was fundamentally an unreasonable position he probably wouldn't do that. But he seems to be running his blog as something of an open(ish) forum for AI commentary. I suspect if he thinks it's a reasonable position that happens to be entirely incorrect he still would post it...

> For example, you say "nobody cares if you reload one page", implying there is a limit -- where is that limit?

You can't demand this degree of specificity from laws - that's not how good laws work. The limit should be reasonableness and intent.

Refreshing hacker news every 5 minutes isn't unreasonable. Refreshing some 5GB file on archive.org that never changes would be unreasonable.

Hell it's easy to come up with examples that would be outright fraud (deceit to acquire something of value) and deserve criminal charges. Refreshing a booking website in a manner to trick it into thinking you're about to reserve something to prevent other people from doing so - e.g. to keep people out of nearby campsites on a camping trip.

> How many kids in a poor country do I need to hire in order to stay under that per-person limit and continue my business as usual?

Up to a jury. But there's clearly a limit at which the answer is "no amount". It's well established that distributed denial of service attacks are illegal after all.

Yeah, but Wright's law (cost of making the n^th something (like a token) being exponentially less than the cost of making the first) strongly suggests that post-bubble inference will be significantly cheaper the longer the bubble lasts.

Whether that's a good thing considering the other things those dollars could be going towards...

No.

Well probably yes but not at all relevant here... the problem wasn't the legitimate extension, it was cursor serving a fake malware one.

Who says there wasn't?

A lot of the most valuable SpaceX IP (at least to the big "advanced persistent threat" hackers) is probably related to military secrets, information to let you spy on/disable/control star shield/link, and so on. Not things we'd find out about as the general public.

While it seems likely... that article actually discusses them probing American waters and air space (edit: actually international air space closest to America), and not Canadian...

I suspect the non-tech-demo version would be multimodal, and include support for textual tool calls...

Converging on what AI companies are already doing... just from the other direction.

I think you're misreading. MPs are directly elected. Anyone can run to be an MP (You have to collect 50 to 100 signatures from supporters and fill out some paperwork). Whoever wins the most votes in a riding becomes MP.

It's the property of a candidate being affiliated with a party - an indirect part of the election - that is run by the parties rules that are generally less than entirely democratic in nature.

I don't have the list of relevant studies I've read on hand atm but there was at least one where they simply pumped in CO2 (or didn't) and had an effect showing pretty directly that CO2 itself is problematic at reasonably low levels - as well as potentially being an excellent proxy for other air quality issues. Problematic in a "makes you drowsy/stupid" type of way - not a "kills you" type of way. Though there's also at least one correlative studies suggesting it might have long term health effects (which really wouldn't be surprising if it has any short term effects).

Besides which, proxy or directly high CO2 means you should ventilate...

CO2 - I'm thrilled with anything less than 800 ppm (half justified by being the lowest number I've ever seen a study suggest might have health effects, half justified by being at a point where I personally don't notice that I'm more drowsy).

PM2.5 - My impression is the health risks here are more cumulative exposure (and include things I'd like to go without in my life, like alzheimers) and a pretty straight line of "lower is better all the way to zero". The WHO's guideline is mean exposure of 5 µg/m³ so I guess I'd adopt that if I had to put a number on it. (But it's a mean, I wouldn't worry about briefly exceeding it)

PM10 - No real opinion (I'd rather worry about PM2.5)

VOC index - I'm not sure this is a standardized thing, but the version I've come across is an index relative to the average that particular sensor has seen in the last 24 hours. So there's no standard possible here.

---

As for a gas range I would worry about NOX considering the studies with them causing and aggravating asthma. That said your CO2 levels are low so I suspect you might have sufficient ventilation that this isn't a huge issue for you (especially if that ventilation is in the form of a fume hood over your stove that you actually use).

Merv-13 means it filters at least 75% of the particles passing through it.

For a standalone air filter which is pushing the filtered air into the same room its drawing from, we can assume reasonably good mixing of the intake and exhaust air. Thus to be equivalent to a HEPA-filter it only needs 1/3rd more airflow... Which really isn't much.

This doesn't follow for a a fancier setup where you're taking dirty air and blowing it into a cleanroom or something... but that's not how we're typically using household air filters.

I believe them. I had Gemini-3.1 Pro do essentially this seven months ago - specifically a loop of writing "Bye! Done!" repeatedly until manually stopped. Presumably some sort of pathological behaviour where it wasn't ranking the stop token highly when it should have been... but damned if I know why... it was an innocent prompt (and it had done the thing I asked it to before failing to stop).

While I'm no fan of that law, it isn't a case of intermediary liability. It's a case of legislators (in a baffling move) deciding that selling 3d printers (and I guess CNC mills?) without vaporware attached should be directly illegal, much like most of the world has decided selling guns should be directly illegal, regardless of how they are used.

It's not illegal to sell these because it would be illegal for the end user to make a gun with them (it might or might not be - but is irrelevant) - it's just illegal to sell them because the legislator said it is.

I think there's at least one clear line between DNS providers and your examples. In your examples the enjoined parted is providing being a tool - a thing someone else uses - not a service where the enjoined party themselves does the thing.

It's one thing to say that you can't CNC machine guns for customers of your CNC-on-demand service, it's another to say you can't sell CNC machines because someone might use them to make guns.

Here the DNS service is being told they can't direct traffic to pirate sites, not that they can't give out tools (e.g. browsers or local DNS resolvers) that could be used by someone else to direct traffic to pirate sites (or could be used innocently). Your examples are all the latter - tools.

Discord does business in the UK, Australia, and just about every other country of any relevance. It would be trivial for the UK courts to collect fines against them simply by ordering the payment processors handling payments for them in the UK to redirect those payments, for instance.

4chan is only reasonably immune from this because it's not really operating as a business...

No, the rotors are tilted by timing the acceleration from the single motor - there's no servo in the design in the video.

I strongly doubt adding a ton of extra weight in aerodynamic surfaces, and drag, and complex mechanical parts, to avoid a few motors/servos is worth it in that context.

I think there's a pretty clear distinction between math and science. The former builds theoretical models of what certain axioms and assumptions imply. The latter builds actual models of what we see and measure in the real world.

Advances in math will probably result in advances in science, which in turn will result in advances in engineering (which is probably where the average person sees an actual benefit in their day to day life). But it's not automatic and a lot of the hand wringing by mathematicians about the AI math results include "is this (AI) method of finding proofs actually going to result in as many of those downstream advances as the old fashioned way?"

(I include theoretical CS as a subfield of math - and believe the S in it is a misnomer).

There's a graph for bugs and bug-fixes (not CVEs, but in linux close) near the end of this article that looks approximately like what you are interested in: https://lwn.net/Articles/1088776/

The use isn't anonymous, it sends your IP address to a third party, a personally identifying piece of information.

This is fundamental feature of all traffic sent via IP unfortunately.

I'd say for the sake of argument we're assuming that's not possible. Perhaps the blood types are incompatible in any direction but killing the healthy person [1]... the point isn't why these are the only options but assuming these are the only options what would be more ethical. Same way the point of the trolley problem isn't to say "derail the trolley" but to probe at which of two options is ethical in the nice simple world where those are the only two options - because if you can't say what's ethical in a simple world you certainly can't in a complicated one.

[1] do you need blood type compatibility for organ transplants? I really don't know... But I'm certain there are some compatibility constraints to ground the example in the real world...

So first off, Asimov's entire point was these laws don't work. But also the entire setup of the trolley problem means that the robots only option is to violate 1. by either pulling or not-pulling the lever (though the server case is different - at least in the philosophical interpretation - I can easily see Asimov justifying it by saying that destroying the train-control-robot is going to cause a lot of human beings to come to harm).

> Just standing by makes total sense. By participating you become part of the problem.

Consider also the variation of the problem where you're a surgeon, you have 5 patients, and one healthy person whose organs/blood/... could save all 5 if only you murdered the one... (but none by simply maiming them, i.e. one needs a heart, one a stomach, one a liver, one so much of this particular kind of blood that you'd have to kill the donor, etc).

> The person at lever would became active participant on that train disaster. All we know the train could go too fast, apderail and kill 100 people.

Standard philosophical interpretations would reject this possibility as impossible (the purpose of the question is to have a simplified environment to probe ethical beliefs - not to reflect reality). I'd also say even outside of that this is like saying you have to consider the possibility that by wearing a green sweater you might distract a driver and cause an accident - you might - but so might doing literally anything else - it's not a foreseeable outcome and has no ethical weight as a result.

Fancy power electronics in them that make sure they only send power to the outlet while the grid is energizing it too.

(And this is intentional to avoid electrocuting line workers who are working to fix the grid when it breaks)

I don't think you're reading it correctly.

I'd describe this as approximately weakening aliasing xor mutability to interior-aliasing xor mutability. The motivation seeming to be to have a more expressive system that rejects fewer correct programs. With a secondary goal of interacting with more (not less) restrictive languages (namely rust).

You can't revoke the old ones, you can choose to only communicate new ones with the restrictions though, and make it other peoples problem to figure out if there's another more permissive license available for that particular work...

For CC By-Sa works the new one might also not be strictly less permissive, because applying the restriction should allow people deriving derivatives to do the same via the share alike clause when they previously couldn't (though I didn't actually check the wording this person is using so this is just assuming the restriction is applied like I think it ought to be).

I tried the wasm version that simonw kindly hosted for them... https://static.simonwillison.net/static/2026/photocraft-web-...

The very first thing I tried to do - create a new document at a specific resolution and DPI - didn't work. It completely (and silently) ignored the values I entered.

It's still impressive in some ways to be sure - but I agree the marketing I've seen for it suggested to me that it is more done that it appears to be. It's the common theme with LLM developed that marketing is pushed far too early in the process.

3 non-formalized papers withdrawn, 6 non-formalized papers formalized. Extrapolating the rate there won't be any non-formalized papers left by the end of November, but because 2/3rds of them will remain as formalized papers.