Posts by diodesign 3533 publicly visible posts • joined 21 Sep 2011
You're in the moderation queue because you appeared (in a few comments) to encourage support for a proscribed terror organization, which makes our UK staff at least uneasy given the legal regime in that country.
Stay on the lawful side, and your comments will make it through. Like your Hamlet.
C.
It is creepy seeing completely empty self driving cars weaving around SF's streets, automated. I saw one following an Uber driver, who quickly pulled to a stop to let someone out. The Waymo SDC in an instant flicked to the right to go around the Uber and continued on. It was very human-like and very slick.
I was impressed. But that was just one instance on a road it must have been trained on countless times. I can see it all working, if conditions are as they were during training. Which isn't a given. There seems to be little initiative in them.
C.
OK, fine: back in the day, as you say, it issued PC GPUs annually. We're writing in the context of today, and specifically Nvidia's response to AI GPU demand. Lately Nvidia has been on a two-year cadence, as I'm sure you know.
I've tweaked that sentence. If you spot something wrong, do let us know via corrections@theregister.com, please.
C.
The letter to Meta is now in the article, with a link. And note: X faces punishment for Hamas-Israel disinfo while Meta could be punished for election-interfering deepfakes.
Meta was also urged to crack down on Hamas-Israel disinformation, but that warning wasn't to the same level as that to X on that front. Subtle difference.
C.
> Didn't notice any mention of multiple platforms in this piece, though
OK here's the deal: the letter to X went out on Tuesday, threatening Musk with fines for allowing this disinfo to spread on Twitter.
Then on Wednesday, our vulture wrote a piece about the X letter. After the article was filed but before it was published, a second letter went out, this time to Meta. That letter to Meta isn't quite the same: it threatens fines over election-related deepfakes. It also reminds Meta to keep a lid on disinfo.
So what we'll do is, update this article with the Meta letter - but it would be wrong to say Meta and X have both been threatened with fines over the Hamas-Israel disinfo, the issue that was the thrust of the story when it was written. X is facing the heat for Mid-East disinfo, and Meta is under fire for deepfakes.
Hope this explains it.
C.
So far the bug(s) are being used in very targeted attacks by commercial spyware. You're likely not a target.
But you want to apply updates as reasonably soon as you can, as the patches are out there, more details will come out, and others may start using the flaws in a wider fashion.
Right now, no, you're unlikely to be hit.
C.
Late to this party but generally the USA says: if you help someone evade our export controls, we'll blacklist or sanction you next. Or mess with your US operations. Or cut you off from the dollar, the world's de facto currency.
So that's why places like ASML play ball. It's not worth the war with Uncle Sam.
C.
I suspect the "regular automatic updates" is doing some sneaky heavy lifting in Google's assertion, but you might well be right. Google may argue its ChromeOS updates over those 10 years exceed what SUSE LE gives you.
Also, "commits to today" -- in the link you provided, SUSE committed to 13 years in 2009, 2014, and 2018.
Not excusing anyone, just commenting. We'll bear it in mind.
C.
We've revised the piece with more details: the flaw is within libwebp, a Google-managed library that processes WebP images.
That same library is present in Firefox, Thunderbird, and other programs. So it's not just Chrome: it's anything using libwebp. Look out for patches and apply them.
C.
That's an interesting thought. But what is equally likely to have happened (IMHO) is that an engineer - a site-reliability engineer, perhaps - was targeted and crash dumps sought out to find bugs in Microsoft products and services that could be exploitable.
Just hunt through dumps that have some kind of memory corruption, and you may find flaws that can be exploited to achieve remote or local code execution rather than a crash.
Or they just got lucky.
Edit: They might also have scoured dumps for sensitive stuff that wasn't redacted -- keys and passwords, etc.
C.
Oi, we wrote about that forking site in 2021, thankyouverymuch.
https://www.theregister.com/2021/08/14/sweary_four_word_map/
It's even in the more-stories list in the middle of the article. Consider yourself tsk'd.
C.
Hey, I think you should give us more credit.
We've extensively reported on OpenAI's failings as well as other LLM makers; we've said in the piece GPT-4 regresses in certain cases; and arguably GPT-4 is more advanced and powerful than GPT-3 - though that doesn't (of course) make either model terribly reliable and safe.
I've made the situation clearer in the piece with a couple of tweaks. Feedback is great, we try to have a thick skin, and we always try to maintain a healthy level of skepticism in articles, but that marketing jibe... ow! :)
C.
It's newsworthy in that it's research accepted by Usenix and vendors have responded to it. Also not everyone is savvy enough to realize the implications of 'don't route LAN connections via tunnel'.
So, yeah, we thought it was a neat little trick and something for people to be aware of. If it was deemed WONTFIX by developers and the attack was too esoteric, we would have ignored it.
And also, we're careful not to say it's the breaking of encryption or anything like that, and that secure connections are still secure.
If it's not news to you, that's great. But that doesn't make it not newsworthy.
C.
Hiya --- FWIW $450 (as in USD) is Samsung's direct price but you can find it discounted elsewhere or part of a plan.
Also we've added in a few more specs for those asking for them. Our 'first look' pieces are more hands-on comment articles about using stuff for the first time as opposed to a detailed review.
C.
Quick look from 2023 YTD, number of unique readers in that time:
Chrome = 52%, Safari = 24%, Firefox and Edge = 6% each. Rest a mix of mobile view on Android, in-app Safari, Opera, etc.
StatCounter has, for June 2023, global browser usage: Chrome = 63%, Safari = 21%, Edge = 5%, Opera = 3%, and Firefox = 3%.
C.
Yeah, don't worry, we'll get to it -- this article was just to quickly flag up that Google has formally gone to the FTC. We've covered Microsoft's licensing pain previously (see the links in the piece) and we hope to uncover all of it eventually.
One of our editors Paul has been writing about this a lot: feel free to drop him an email <pkunert@theregister.com> and let him know more. Cheers!
C.
Yeah, the end of those cookies will be like that moment in the Titanic movie when the boat splits in half and tips up. There's those who manage to hang on and those who fall.
It's gonna be like that for publications IMHO. We'll see who prepared for it and who didn't. It's something we've been working on for a while.
C.
One interesting aspect of this - speaking generally about the online publishing world - is that no one believes each other.
If a vendor goes direct to a B2B publication, books an ad campaign, pays $20,000 for 300,000 ad impressions, and then after a few days or weeks, the publication says, 'ta-da, all done, thanks for the money,' that vendor may be skeptical the ads were actually all served to real humans. Especially when the ads are targeted and we're in an era of ad-blocking.
So the advertiser wants in on the analytics to verify the campaign was truly fulfilled and/or to see it progress in real time. The ads in that case end up being served by or involve a third-party both the advertiser and the publication trust / willing to use.
I'm simplifying here to avoid talking about specific operations, and we're happy with our ad clients, and our ad clients are happy with us, of course. My point is, the industry is a very very long way from placing a banner ad on a page and letting it run its course. It's more complex than most may think.
C.
