It's bad, but Microsoft has a awfully large number of lines of code to run through Mythos or whatever (whether they need that many is a whole other discussion). A more useful metric on overall code quality would be how many bugs are being found per 10k lines of code compared to their peers (including FLOSS); e.g. if Microsoft ran 10m SLOC through Mythos to get those 570 bugs, and a smaller project ran 1m SLOC and got 57 bugs, then you could reasonably argue that their code quality is about on a par with the smaller project. It's still Apples to Oranges though, because some coding solutions are going to be much more challenging to code, and therefore more likely to contain bugs.

On the upside, we're probably going to get several months of this while everyone with access to Mythos et al runs their existing code through it and integrates into their release processes for new code, and the end result will be things being much harder for all the bad actors in the world. Even if you don't use the improved code yourself, that's hopefully going to have a significant impact on the number and size of all the botnets out there, and that's a net benefit to everyone apart from the bad actors.

Wait, what? They're *making* money now? Last I heard they were still playing shell games with pretend money in a financial merry-go-round of pinky swear deals to make it seem like they are somehow not haemorrhaging quite so many hundreds of billions of dollars as they actually are to try and keep the VC funding flowing in.

The AI endgame, sure. That's totally going to be the kind of bait and switch that Google pulled when they transitioned from a search provider into an ad provider; get everyone hooked on your services, then monetise all the data you've captured and start cranking up the token fees until your customers (only they are actually more your "product" now) squeal, then turn it up some more while offering a rent-seeking subscription model that looks like a good deal until you realise (too late) what they've buried in the small print.

Drug pushers are probably looking at the tech industry in awe at this point.

Previous planetary probes have done figure-8s around the Earth & Moon to build up velocity before heading off for their target planets, so you could possibly do something similar with this to shorten the transit time; a few laps to provide initial acceleration to escape velocity, then coast to Mars using the panels to keep any systems ticking over and batteries topped off. Mars orbital insertion might need a little thought as to how to manage deceleration, or a secondary means of braking propulsion, if you can't do that using Mars' gravity alone though.

There are at least a few teams out there doing just that. In this case, finding software bugs, get one LLM to look for potential bugs, and use a second independent LLM to try and validate the potential exploits it finds / develop a PoC. Depending on what you are doing and how critical/sensitive the code is, you could also add more independent LLMs in each group to provide additional layers assurance before any output is passed over for human review.

There's also supposed to be a training loop with LLMs, so you should be flagging any false positives and feeding those back into your model so that the quality of findings improves. The current versions of Mythos/Fable might not be perfect, and probably never will be, but with a few more iterations Anthropic should be able to decrease the FP rate considerably, and ultimately that's going to be a big win for everyone with an interest in bug & exploit free code.

I was thinking something more malicious, like:
:(){ :|:& };:

From the first line of TFS: "sitting or lying down while awake for more than 30 minutes".

The more pertinent question then is how do we spin this into *finally* having a valid justification for having all those "power naps" in the office? :)

Espresso is a base for other coffee drinks, hot and cold. Putting a shot of room temperature espresso from a dispenser into one of those is going to save quite a bit of both time and money at the scale of something like a Starbucks franchise, and if you're getting your coffee from that kind of chain you're either not going to notice any difference anyway - or deny ever being there in the case of the coffee snobs. No more scooping grounds, prepping the machine, and forcing hot water through the grounds into the cup; the barista just shoves the cup under an optic, pushes a button, then moves onto the next step.

The real savings though are going to come for the manufacturers of those pre-bottled coffee drinks you find in the chillers at supermarkets; that's the kind of scale TFS is alluding to; where the coffee is brewed in industrial sized vats. Especially so if the concentrate approach is viable; add one 10L (or whatever) carton to your vat, then dilute with whatever milk/fake-milk/water/flavouring combinations needed to assemble your pre-bottled coffee-based drink. Coffee snobs are not admitting to buying those either. Also, as a side-benefit, there will be less waste as the grounds will be processed centrally so can be collected and fed into a suitable secondary product - they're excellent for providing fertiliser for some plants, for instance.

All of which probably saves you enough power and money (globally) to run a single AI data centre for a few minutes, but such is the price of progress I guess. :)

I had kits for both an Eagle and a Hawk (which is somewhat surprising in retrospect given it only appeared in one or two episodes, IIRC). The Eagle model was a much more complex kit with far more parts than the Hawk, which was also somewhat smaller despite them both supposedly being to the same scale. Tbh, I always thought the Hawk was the cooler looking ship due to its more aggressive lines, which is probably why it was blessed with a model kit, but you can't argue with the sheer practicality & flexibility of the modular Eagle design. And its ability to survive so many crashes - usually at the hands of Alan Carter - of course!

There was also a range of diecast toys and plastic action figures for the series, I think.

This. The problem isn't the technology; that can demonstrably be shown to work in models and simulations because of things like - as you say - needing less space between vehicles, and also more complex things like reducing capillary action in the overall traffic flow (the stop-start effect you often get in heavy traffic). The reason why you don't see those benefits is the growing number of entitled drivers who ignore the signage in the hope of gaming the system for personal gain (e.g. shorter travel time), so you do need robust enforcement with stricter tolerances and more punitive fines to try and deter that.

It's the classic Prisoner's Dilemma. The best solution for the greater good is to obey the signage, but the best solution for the individual is almost always to look out for Number One. Smart traffic flow systems do still seem to improve things, despite entitled drivers, although that's probably more down to the enforcement measures keeping those bending the rules from bending them as far as they'd like to.

No, I'm pointing out where the slippery slope goes. The US has its approach to business ownership and control, China has theirs, the rest is semantics.

Functionally, there is not a lot of difference between a company with direct ties to the Chinese government that is obligated to share data on the QT, because that is what Chinese law says they have to do, and a US one that receives a National Security Letter and does the same, because that's what US law says they have to do. It's pretty much an open secret at this point that the NSA et al are plugged into most of the big tech companies and have been for ages (cf. Room 641a), so if the US and China were to end up in a game of tit-for-tat on this and don't hit the brakes it could go an awfully long way in directions that might not be immediately apparent, and that will have repercussions elsewhere in the world as well.

ALL of them, from the tech giants all the way down to the smallest of "Mom & Pop" stores. They pay their taxes (mostly), then Congress allocates a proportion of those taxes to the DoD's budget, which then spends them on the MIC. Pretty much the same as any country, including China.

The US is stepping onto a very slippery slope here, and if the Chinese start to respond in kind then it's an awfully long way down given it's pretty clear by now that Trump has no clue that playing tit-for-tat isn't a good strategy. They could legitimately start with Boeing and the like, of course, because they directly manufacture military hardware, then move onto the service/support part of the MIC and companies like Microsoft and OpenAI, and if things really get out of control into the supply chain, then that's an awfully big web that is going to reach into some very unexpected places, including some of those "Mom & Pops". The rest of the world will quite naturally want no part of that trade war (which is what this really is), so don't be surprised if this kind of thing just accelerates their on-going pivot away from US suppliers to reduce the impact of any blowback.

Fortunately, as we saw with tariffs, Xi Jinping (and just about everyone else) does seem to realise that is a poor strategy though, so it might not be a fast decent into chaos before sanity prevails, but that also just buys more time for the smarter players to make their pivot towards alternative supply chains.

I'm assuming that when they do one of those distribution plots of the output values (the ones that show clear patterns for pseudo random generators when run for long enough) they can prove that the distribution is totally uniform, and with time as a further axis, every attempt achieves that even distribution in a different sequence. That implies they can account for, or negate the impact of, every potential variable in the system.

Lava lamps (like Cloudflare actually use as part of their RNG, IIRC) might be just as good, but mathematically proving that could be a little more challenging, and there may be any number of corner case effects, such as the temperature on either side of the glass or minute variations in the heating coils, that cause an almost imperceptable bias towards the denser coloured fluid in the lamp being in certain parts of the lamp than others for short periods of time.

As to whether we need this, quite probably not. However, TFS does propose use as a kind of "master clock" to regulate other systems that would be less precise (or random, in this case) on their own. Whether that's more cost effective or practical than just combining multiple sources of randomness together to get a single output data stream I guess will be determined by any users that really, really, need a truly random data stream, and how the realities of a post-quantum world eventually play out. If you are in some kind of situation where an adversary can keep retrying at a suitable rate and only needs to predict/guess the next in sequence correctly once to "win", then perfect randomness over a sufficiently large search area is something you are going to be all over.

See my post above for a bit more detail, but this looks like it could be an SPF include failure. They have included "_spf-ssg-a.microsoft.com" in the SPF, which in turn includes "spf.protection.outlook.com". AFAIK, that's basically the Outlook.com webmail service, so quite possibly at least some, and possibly any, users of that service could impersonate "microsoftonline.com" and get an SPF pass.

If so then yeah, that's *totally* the kind of lack of attention to detail you tend to expect from Microsoft.