How is that relative? It literally won't do anything unless you trigger it to generate the preview. If you've chosen to generate a preview of a link, you've chosen to visit that link, and that is entirely on you.

It's not even that. It's a simple web page FizzBuzz test (print the numbers 1-100 on a web page when you click a button and do FizzBuzz-like things). We give them a day to do it and tell them that we will review what they write in the technical interview. People create elaborate solutions to impress, but I've noticed certain... trends in how they design it. And after the review, when we ask them to make it so the FizzBuzz factors are user editable on the web page, things always end up falling apart.

We provide a technical test in our interview process and we got all sorts of neat submissions but it always falls apart when we ask them to make changes live. I personally don't care much about somebody using an LLM if they can demonstrate that they are skilled enough without using one, and if they can demonstrate that they won't trust a result. In fact, at this point we ask about AI usage and have applicants ask it questions that we know will give wrong answers just to see if they can identify the problem and figure out how to resolve it.

Well, you can choose to just not install it. It doesn't ship with the browser and you have to enable it and download the AI model explicitly. But, you can go to "Tools > Extensions and Themes > On-device AI" and delete any AI model that you may have installed if you don't like it.

I mean, it doesn't even get installed until you choose to install it, and it can be removed at any time. I don't think having AI extensions that aren't shipped with the browser and need to be explicitly enabled don't really qualify as "bloat". This should just be an article about how their smart tab grouping feature is broken.

They make it clear when you install the add-on. It explains your highlighted text will be collected and sent and it requires you to opt-in before it can be used.

I would agree with that. It's anecdotal but I've noticed when using Copilot at my job that it usually gets me a "mostly" proper solution. But even getting you mostly to a solution can save you an hour or more of digging through documentation. "Hey Copilot, I have an Excel workbook in a memory stream. Load it up with the Open XML library, open up the Summary spreadsheet, and copy out the contents of cell D:3." AI bots are pretty good at crawling through lots of information and summarizing it; I've been finding them quite good at identifying which classes and functions you need to use in unfamiliar libraries. But, in the end, you still need a developer with enough skill to understand the provided solution and know where the solution was deficient, since, as I said, it usually just gets you part-way there.

They claim about 20% more battery life over the Pixel 8 when the screen is on. The 8 was pretty good already so I'm hopeful, but we'll just have to wait for reviews of the device.

The Mozilla CEO said that was a UI issue that they will fix in a newer version of Firefox. Telemetry being disabled does prevent the feature from being used.
https://bugzilla.mozilla.org/s...

The UI fix should be hitting beta, nightly, and 128.1esr.

It looks like Citra (3DS emulator) may have been managed by the same entity so Citra has been removed too.

I really doubt Apple wants to get into the regulated healthcare industry and be on the hook for all that hospital tech Masimo sells.

Firefox extensions can do things that Chromium based browsers can't do. There are also some APIs that Firefox doesn't support.
https://developer.mozilla.org/...

Sure, you can't have an extension rewriting the browser UI anymore, but there is still a lot that extensions can do. And you can always re-style the UI on your own with UserChrome.css to get that Windows 95 look back.

Yes. I've seen people who work on the filters say that YouTube is updating its scripts about twice a day. The uBlock Quick Fixes filter auto-updates every 12 hours (and possibly every 6 hours once .patch filters are implemented). This page has been keeping track of uBlock's ability to bypass YouTube ads: https://drhyperion451.github.i...

Basically, once YouTube updates the script to bypass ad-blockers, uBlock will shorty follow with a fix. Unfortunately, it isn't instant, and you may have to manually update your Quick Fixes list. It's an arms race.

I can't really blame Mozilla that much. They were using OS functions normally in an intended way and there was no performance problem with the way they called those functions. They tried to get Microsoft's help and attention 6 years ago but they were literally just told to post about it on the Feedback Hub and hope Microsoft even bothered to investigate it. Looking at the ticket, it looks like it took somebody basically reverse engineering Defender to figure out what was going wrong before Microsoft acted on it. And yeah, they probably expected it to just be an issue with Defender. It wouldn't be the first time an antivirus misbehaved. And honestly, it IS mainly just an issue with Defender. The optimizations Mozilla is making now won't really improve anything with Firefox. It will just make OTHER applications on the computer behave a little better.

No, Chrome and Edge just use VirtualAlloc, which antiviruses check differently. The problem was that when Defender saw Firefox make a call to VirtualProtect, it tried to parse the ETL event. This parsing function was written poorly and ended up allocating and de-allocating a 64KB zeroed memory block for each of the 18 different event attributes. So every single time Firefox called VirtualProtect, Defender would basically malloc/free a 64KB memory block 18 times. And frequent memory allocations are not cheap.