AI Harms Open Source Projects, Android Goes Private  ## Metadata - Author: [[Linux & Open Source News]] - Full Title: AI Harms Open Source Projects, Android Goes Private - Category: #podcasts - URL: https://share.snipd.com/episode/adc7e029-584a-43c8-ad8f-ee55739af229 ## Document Notes ## Highlights - **AI Bots Impacting Open Source** - AI bots strain open-source projects by increasing bandwidth costs. - This happens without compensation from AI companies, causing instability and extra work. Transcript: Nick And this week, we've got a lot to talk about. First, some open source projects are really starting to feel the pressure of AI bots trying to scrap all of the code that they have, and this is increasing their bandwidth costs without, Of course, AI companies offering any form of compensation. We also have Android moving the entire development process to a private development process, purely internal at Google, although the project is still open source. We've got the release of the kernel 6.14, we've got HP giving indications that SteamOS might be their preferred choice if they ever made a handheld, we've got the beta of Ubuntu 25.04, And a lot more. So as always, if you want to know more about any of these topics, all the links that I used to write this show are in the show notes, all the links to help me make this show and to support it Are in the show notes as well, including a link to our sponsor Tuxedo Computers. You all know about them by now. They are a maker of computers that run Linux out of the box. If you need a new computer, whether it's a laptop, a desktop, or even a two-in now, because they have a touchscreen PC now, you can go to their website and you can customize every device With a nice choice of components. You can have your own logo engraved on the lid. You can have your own keyboard layout if you prefer. They are based in Germany, but they ship to most countries in the world. And I personally only use their devices these days. They're really, really good. I've been running my entire YouTube channel and this podcast from a Tuxedo computer for more than a few years now and they've been sponsoring the channel for more than three years. They're really good. The link, as always, is in the description. Go check them out. So not only are AI tools detrimental to people's actual intellectual faculties and reasoning abilities, not only does it consume stupid amounts of resources and energy to accomplish Stuff that we could really dispense with, but it also is bringing some parts of the internet to its knees thanks to the associated crawling bots that it uses. Technically, you should be able to block these bots from crawling your website using the actual robots.txt file, but this file is not followed by most AI bots it turns out. You can try blocking them based on the user agent or by filtering traffic, but generally it just doesn't work. These bots adapt and change how they appear so they can ignore whatever barrier you put in place to keep ownership of your content or your code. A recent report shows that 97% of the traffic that a bunch of open source projects see is generated by AI bots, which means much higher bandwidth costs, instability for everyone else Trying to access the servers because they are being completely saturated stupid requests from stupid bots, and it's generally more work for the open source developers or maintainers, Work that is not being paid for by the AI companies, of course. Some projects had to block entire countries to solve the issue, and some, like Gnome's GitLab, had to implement a form of AI maze, so the bots had to solve complex problems before they Could access anything. They tend to fail these challenges because, well, they're bots. Only 3.2% of the requests on Gnome's GitLab passed these challenges, apparently. So yeah, that's 96.8% bot traffic. KD's GitLab was even kicked offline at some point because of a bot originating from, apparently, Alibaba. The situation has been referred to as an internet-wide DDoS attack and the costs for fast projects have risen enormously. And if they don't want to pay for these increased costs, they have to take time away from developing their cool software to deal with that problem, which is not what anyone wants to do With their valuable spare time. So you've got these AI startups raising billions by investors and at the same time they're increasing the costs of virtually everything else that their bots are crawling. It's not just open source projects. It's everyone's website. It's everybody who pays by the bandwidth. This is not okay. These companies do not follow any rule established on the internet. They do not follow copyright, licensing, attribution. They try to cram everything into fair use. Information wants to be free as proven with how they treat open source projects, open source Git labs. This is BS. They just do not want to pay to access the content, but they want to make customers pay to access a paraphrased form of that content that they've stolen off of everyone else. This is not okay. We really need laws to restrict how these companies can work because I'm sure that if they had to pay, they would all close down shop. Their business model is just non-existent. ([Time 0:00:19](https://share.snipd.com/snip/ee661a6e-551b-45b8-b173-ea169b2f958b)) - **Android Development Goes Private** - Android development is moving to a private process, limiting community involvement. - While the code remains open-source, real-time development visibility is lost, potentially impacting custom ROM development. Transcript: Nick It's unsustainable. Now, it looks like Android's open nature might be in trouble. They have confirmed to the Android Authority website that all development of Android would now happen fully in private. Now, the code will still be open source, just like it is right now. Currently, Google has two branches. There's one branch, which is the Android open source project or AOSP, and this is every commit, everything that is fully open source and fully licensed. It accepts third-party contributions. People can try and submit patches, even though Google obviously decides what goes in or not at the end of the day. It's not a community-driven project. Google also has a second branch, which is the internal dev branch. This one is not open, it is not open source, and it's only available to companies that have a Google Services licensing agreement, which places restrictions on what they can do with That code and where they can talk about it or publish it. Google will drop that current process on Android. They will move to fully internal development work, meaning that you won't see all the patches, all the comments, everything. You will just see the final source code for a full Android release. They say it's to simplify development and to avoid merge conflicts between the two branches, but the next result is no one will have a window into Android's dev process anymore. It is still open source. You still get the resulting releases once the internal dev is done. You will get a snapshot of the code of each major or minor release, but there will no longer be a work in progress branch that is visible and accessible to anyone during development. Weren't interacting with the source code, you won't care that much, but the companies that used to build custom ROMs from work in progress in the Android public branch, they will now Have to wait for a full major or minor release to be out to start their work on porting their ROM and their applications and their customizations to that new Android version, instead Of being able to anticipate that work by taking an almost finished branch and starting that work from there. So this might result in even longer times before custom ROMs move to a recent Android version. This also means that anyone contributing to the Android open source project will basically no longer be able to because all your contributions will be made against an already released Stable version instead of a work in progress, meaning there are higher chances that your patch just solves nothing or doesn't work with the new version and will just not be accepted. I am definitely not a fan of this solution. I would have much prefer them to go the opposite way, which is we're closing down the internal dev branch and we're doing everything in the open. And if companies want to develop stuff that is not fully open source, they can do it on their own servers, not our internal branch. I think this would have been better for everyone. This is definitely not a good move for all the Android open source communities. It's still open source, but it's still not a good signal. ([Time 0:05:13](https://share.snipd.com/snip/270f7d2e-37eb-492f-a361-4f744fa6a121)) - **Linux Kernel 6.14 Released** - Linux Kernel 6.14 includes gaming improvements like NTSync, potentially boosting Wine/Proton performance. - Other enhancements include GPU workload protection, scheduler improvements, and better disk writing. Transcript: Nick Now, the Linux kernel 6.14 was released this week, bringing with it some solid gaming improvements, notably the addition of the NTSync driver. This is something we talked about at length for the year or so that will bring F-Sync or E-Sync to the kernel itself with better performance and fewer bottlenecks because this implementation Runs at the kernel level instead of the user space level. Wine and Proton will need to be patched to implement support for that for this change to actually matter and if you were not using fsync or esync so if you weren't using Proton but you were Using Wine, for example, you will very likely boost FPS by 50 to 150%. But if you were using Proton or a version of Wine that uses F-Sync or E-Sync, chances are you're not going to see many differences. The Proton developers already said they wouldn't rush to implement support for it. They already use F-Sync and they say this should be at least as fast. There's also GPU workload protection in the kernel now, which should better count and attribute the actual memory being used by various processors groups using the GPU. So these processors won't be terminated improperly by the kernel because the kernel thinks it's using too much memory. There are also some performance improvements to the scheduler, so the thing that puts various processes on the right CPU cores, and also improvements on writing data to disk. Intel GPUs will get improved drivers, whether they are discrete GPUs or integrated ones. AMD got some drivers for their latest NPUs, AMD CPUs should use much less battery and much less power by default by defaulting to balanced performance mode and of course you can still Move to the full performance mode if you need it. The kernel now supports the copilot key that Microsoft makes hardware manufacturers add to their PCs and you'll be able to map that key to whatever the hell you want. I guess distros will find a way to use it as a modifier or to make it usable by default. There's a lot more gamepad drivers in there as well, there are some new rust bindings and a lot more. Now chances are the gaming improvements have been overblown quite a bit by a few websites. It looks like for Proton users it's really not going to change much even after Proton actually supports the NTSync driver. Some people reported that in certain games NTSync does improve FPS over Fsync but it doesn't seem to be a general big improvement. If you were only gaming with Wine without F-Sync or E-Sync though, so maybe you were using a custom implementation of Wine on the Heroic Games launcher or on Lutris, you might get much Better performance with that if those custom Wine implementations use anti-sync instead. ([Time 0:08:32](https://share.snipd.com/snip/0b599155-d4dd-4f54-a948-da542e84ad05)) - **HP Favors SteamOS for Handhelds** - HP may favor SteamOS for handhelds, citing struggles with the Windows experience. - This could lead to more manufacturers supporting SteamOS, offering a simpler gaming setup. Transcript: Nick If you're not concerned with all that Linux gaming stuff though, you will get much better battery life and performance technically on amd cpus and amd laptops which is good and you just Get more drivers and better hardware support which is always a good thing now hp seems set to use steam os instead of windows on potential gaming handhelds the president of gaming solutions At hp said in an interview for XDA developers that they struggle with the Windows experience themselves. They don't like it, and they don't know how to make a good product around that Windows experience. When asked if they would be interested in developing a SteamOS-powered device, they said yes, and that if they're buying a gaming handheld, they just want a very simple setup. For example, just a device being able to remember the last game they played and putting that first and foremost, which Windows obviously doesn't do because it doesn't have a dedicated Gaming interface. You need to develop your own for it as a manufacturer and plenty of manufacturers do, but HP doesn't seem inclined to do it. HP does have their Omen gaming brand, so maybe at some point we'll see them bring a handheld running SteamOS, which would make two big name manufacturers supporting that OS, HP and Lenovo. Wouldn't be too shabby. Of course, it could also just be a warning shot fired in an interview at Microsoft to see if they plan on improving the experience there. They said they were working on it, but for now we've seen really nothing. Of course, until there's an actual HP Omen gaming product on the shelves, it doesn't matter, and even if that happens, that gaming product needs to be good. If it's just the less powerful, also-ran version of a Windows handheld, then it doesn't help anyone. If it's like the flagship running SteamOS, as in we're not offering anything else, you just have the SteamOS version because that's the only one that is actually usable, then it is probably Pretty nice. I think Lenovo's strategy is not too bad, they had the very high-end Windows one and the entry-priced SteamOS one, except the SteamOS one also can go up to the flagship edition and it Still costs a little bit less money, so I think pricing-wise and entry-level it makes a lot of sense to do it like Lenovo and their Legion Go S. ([Time 0:11:23](https://share.snipd.com/snip/ad6dc51f-f159-45ab-80fd-8cfd9fd6b43a)) - **Ubuntu 25.04 Beta** - Ubuntu 25.04 beta features kernel 6.14, new PDF reader 'Paper', and installer improvements. - It also includes theme enhancements, symbolic icons, and Nvidia Dynamic Boost support. Transcript: Nick Now Ubuntu's latest beta is out, namely for 25.04, which will release in less than a month. This new version updates all the internals as usual, and in keeping with their recent kernel policy, they will use the kernel 6.14 that just released. We already discussed those at length, I have a dedicated video. So there's HDR support, performance improvements, the well-being settings, battery charge limiting, updates to a lot of core apps, better support for Wayland protocols, and the Global keyboard shortcuts portal. As per Ubuntu specific changes, there's the replacement of the EVENTS PDF reader by Paper. It is a newer GTK4 Libet Vita app partially rewritten in Rust from the events codebase and it looks much more at home on modern GNOME. Likely it will at some point become the default PDF reader for GNOME entirely. You also get an improved Ubuntu installer that better handles dual boots with Windows, if your windows partition is encrypted using bitlocker you also get a new api to power location Based features notably nightlight there's also some theme improvements to the icons the rounded corners squares that ubuntu uses will look much nicer because they've been scaled Down a little bit so they are just better with the other icons. There are also some more symbolic icons all around and support for Nvidia Dynamic Boost. That's a feature that lets your hybrid laptop use more power for the GPU when it needs to by lowering the power allocated to the CPU if it doesn't need as much. This is a pretty useful feature for gaming because some games depend a lot more on your GPU than on your CPU, they're called GPU bound, and so in that scenario technically this thing could Help a lot. Now the beta is out right now, you can download it and give it a shot, you can also expect a dedicated video in April on the YouTube channel to talk about this. There are no stated improvements to the Snap ecosystem, the permissions they introduced, or Snap packages themselves, or the loading times, or the open up times, or how the Snaps connect To each other, the various APIs they support. So if you didn't like Ubuntu because you didn't like Snaps, this probably will not fix that. But if you weren't annoyed at Ubuntu for their snaps and you felt the snap ecosystem was fine, then I think this is a worthwhile upgrade on paper at least. ([Time 0:13:46](https://share.snipd.com/snip/6e7004fe-9994-4ad0-a060-cf9818eed33e)) - **GRUB and Rust** - GRUB gets a Rust code prototype, potentially addressing memory management vulnerabilities. - Despite some opposition, Rust's technical merits in enhancing security are well-established. Transcript: Nick We'll see when I can try it out fully on a real computer. Now, opponents to Rust will have one more thing to complain about, I guess, because Grub might be getting some rust code in as well. The Linux kernel got some rust drivers, Ubuntu is replacing some Korg new utilities with rust-based alternatives, and Grub looks to be following suit. This is just a first prototype for the i386 platform, it doesn't really work entirely yet, it doesn't handle Grub panics gracefully, it doesn't support all architectures, so it's Not going to turn Grub into a fully-fledged full Rust project overnight, but as expected, some people are not happy about this thing even existing, saying that the technical merits Haven't been discussed and that this makes it feel like Rust is used to show off instead of being used to solve problems. Most of the feedback is being reported as positive though probably because Grub is full of vulnerabilities linked to memory management and that's the exact purpose that Rust was created To solve. And honestly, that technical merits argument pops up every time someone makes something in Rust, and I think it's been discussed to death, basically. It's the only thing that is ever talked about when anyone does anything in Rust, is, is this thing technically relevant? Why are you using Rust for this? And we've seen all the justifications that are always the same. It's faster to write code for it, there are less vulnerabilities at least linked to memory management. That's it. That's what people have been talking about for 7 years. So saying the technical merits haven't been discussed seems pretty weird to me because we all know, even non-developers know what the technical merits of Rust is at that point. ([Time 0:16:19](https://share.snipd.com/snip/afdad928-696c-41f1-91b5-2bbda372c0e0)) - **France's Digital Suite** - France launched "Suite Numérique," an open-source alternative to Big Tech services. - Initially for public servants, it includes chat, file transfer, spreadsheets, and video conferencing tools. Transcript: Nick Now, France has launched what they call the Suite Numérique or Digital Suite, which is an alternative to the various online services offered by Google and other big tech companies. It looks like it is entirely open source, but it also looks like, for now, it is limited to public servants working in local communities, in town halls, in regional administrations. It is not open to the general public right now. It contains a bunch of tools. First, there's CHAP, which is probably short for chat app. It is a Slack slash Microsoft Teams equivalent that sort of looks like Mattermost. There's friends transferred, which is a large file transfer solution. There's grist, the spreadsheet module. There's also docs, a simple collaborative text editor that doesn't seem to have a ton of features yet. There's visio, which is a video chat that really reminds me of Jitsi Meet. There's messagerie for messenger. It is a webmail made for civil servants as well. You can connect to the entire suite with one single login, it is all packaged and accessible as a single suite, and apparently that login can be used by other providers to add tools to The suite and store all the relevant data in a single account. All that data is stored in France, of course, and the goal is mainly to equip civil servants, not the general public yet. Now, it does look interesting. It doesn't look in any way remotely comparable to the full digital suites you could get from, I don't know, Microsoft 365, or even Google Docs, but it is an interesting initiative trying To, let's say, de-big tech European or at least French administrations. And I could see a future where that suite gets bigger, gets better, gets some actual investment from the government or from European funds, and becomes the de facto thing that you teach Students and that they learn to use, that companies rely upon. Would be pretty neat if that happened. That would be one more alternative that is fully open source to all the tools of, well, big tech companies. I probably wouldn't trust a login handled by the French government because that's a government. I don't trust any of those. They could always turn bad and they have turned bad in the past. But yeah, I think it would be nice because it is open source, so anyone could host it and it could become pretty cool. ([Time 0:18:12](https://share.snipd.com/snip/dd95776b-144c-426d-8a52-e0f821976163)) - **KDE Proposes SDDM Replacement** - KDE developers propose replacing SDDM with a Plasma-integrated login manager. - This aims to improve theming, functionality, and address existing SDDM limitations. Transcript: Nick Now, KDE developers proposed a change to replace SDDM, their current login manager. It's not really currently a KDE or Plasma project, but it's been used as the default Plasma login screen for ages. And apparently, it still has limitations that make improving it more difficult than the Plasma team would like. SDDM is also meant to work with any desktop environment. It's not meant specifically to be a Plasma project and this also seems to be creating issues. So the proposal would be to create what is basically a mini Plasma session so developers could rely on all of KDE's existing APIs and subsystems without having to reinvent everything. It would also allow this login screen to be themed and handled like a normal Plasma workspace, and they would use the SDDM backend still as the base to handle all the logins, logouts, And stuff like that. They've created two new repos, the Plasma Login Manager, which is basically SDDM's backend but modified specifically for KDE, and there's Plasma Login, the frontend to replace the Graphical part of SDDM. They say that the code is near feature parity already, but that it is too early to be shipped by any distro. Which probably means it's too early to be shipped in Plasma 6.4 in June, but it might make its way into Plasma 6.5. Of course, the login manager is handled distro by distro, but if Plasma recommends their own thing it's generally what distros will ship if they offer a specific Plasma-focused version. And I've been noticing a bunch of incoherencies and problems with SDDM, it doesn't retain the keyboard layouts that you've set up even as administrator, it doesn't retain the language Choices, some of the scaling is not right, there are a bunch of issues here and there, so if this new login manager can fix that, that would be pretty cool. ([Time 0:20:37](https://share.snipd.com/snip/5d989e61-7a40-4162-9e1b-55cab404392f)) - **Zorin OS Switches to Brave** - Zorin OS 17.3 switched from Firefox to a modified Brave, disabling many features. - This move, citing ethical concerns, raises questions about Brave's suitability. Transcript: Nick Now there are some changes happening in web browsers in the Linux and open source world. The first one is Zorin OS, they just released their new version of their distro, it is a GNOME based distro, Ubuntu 22.04 distro, so pretty old base, but they released 17.3. It doesn't contain a lot of really important changes, but what they did is move from Firefox to Brave. They're saying that Firefox and Mozilla in general no longer align with Zorin's values. They're probably referring to the recent Terms of Use controversy. And so they've moved to Brave, they've adjusted the UI to look more at home on Zorin, and they've disabled a lot of Brave's features like the wallet, Brave rewards, the Leo AI assistant, Sponsored background images, and a few others. Which is both good and bad. It is good because these features are, for a lot of people just bloat and probably shouldn't be part of a web browser by default but it's also not so good because then Zorin OS is shipping Something that has the Brave logo and is called Brave but it's not really the full Brave experience. They're shipping a heavily modified version of it to the point that someone expecting the full version of Brave will just not get that and might be a bit annoyed. This move also seems to gloss over the various controversies linked to Brave as a browser or a company that, at least in my opinion, are a lot worse than what ails Firefox and Mozilla right Now. First, there's all the crypto BS where Brave actually recommended some disreputable coin exchanges like FTX for a full year, FTX being one of the worst crypto scams to have ever existed. There's the fact that Brave developers at some point planned to replace ads from websites with their own ads, directly stealing revenue from everyone having a website on the internet. There's also the fact that they sell their search engine data to AI companies, and this includes copyrighted works, because their search engine grabs text snippets from websites, And those snippets are part of what Brave sells to AI companies. So they're selling content that isn't theirs to sell. And that's not counting all the shitty stuff that the CEO and founder has done against equal rights or even spreading COVID misinformation. So this move from Firefox to Brave, citing safety and trust as the reasons, feels like jumping from a tiger's back to a shark-infested pool. You're not really trading up here. I understand why people would not want to use Firefox or Mozilla products anymore. I personally moved to a Firefox fork because I feel that Mozilla is being very disingenuous right now. But Brave is factually not a more ethical or respectful browser than Firefox right now. It is personally, I think, at least worse, a lot worse than Firefox. And the fact that they had to remove a lot of the features from that browser to actually make it function in a way they want is very telling. So don't really think this was the right move. They probably should have moved to something like Chromium with extensions patched in or just to a Firefox fork. Brave is not the solution if your concerns are ethics. ([Time 0:22:34](https://share.snipd.com/snip/a6f1a360-2aa1-4dca-ba4e-0797c992a6ee)) - **ProtonVPN and Vivaldi Partnership** - ProtonVPN partners with Vivaldi to integrate VPN directly into the browser. - This partnership aims to enhance privacy, although Vivaldi isn't fully open-source. Transcript: Nick Another move that we've seen this week in the browser space is ProtonVPN partnering up with Vivaldi, an unfortunately not fully open source browser. The partnership is to integrate ProtonVPN straight in the browser, the rational being that want to stop Big Tech's data collection and that a VPN can help a little bit with that and it's Also having a partnership between European companies because Vivaldi is based in Norway and Proton is based in Switzerland. I'm not sure why Proton would decide to partner with a non-fully open source browser but I've only heard good things about Vivaldi and their closed source part is only their UI. It is apparently auditable, or at least you can take a look at it even though it's not open source, and no one has reported anything shady. Still, it's not like ProtonVPN is going closed source, they're just partnering up with a non-fully open source browser. Doesn't really hurt me. As per the Brave thing, anyone using Brave and liking it, that's fine, I'm not judging you, you use whatever the hell works for you, I just think that citing security or concerns around The ethics of Mozilla to move to a browser that has been proven to be pretty disreputable is not a solid move. ([Time 0:25:53](https://share.snipd.com/snip/5b3e75cc-8ce4-455c-98ed-5ca6a3456807)) - **Nova NVIDIA Driver Update** - The Nova NVIDIA driver, written in Rust, is progressing toward integration into kernel 6.15. - This is a significant test for Rust's capabilities in kernel development. Transcript: Nick Those are not the reasons you want to move from one to the other, I guess. And to conclude on today's episode, the Nova NVIDIA driver got more love this week as it's now set to be merged for the kernel 6.15, so the next release of the Linux kernel. This driver is currently just in a stub form, meaning it's a skeleton of the code. All the necessary features are mapped inside of the driver's code, but they do return fixed values instead of actually interacting with the GPU itself. This is starting to change though, there are about 300 more lines of Rust code landing in that driver that will create a DRM device, so a device that the kernel's direct rendering manager Can recognize and map in the system, and it also creates the first basic input and output controls to create something that can access the actual VRAM of the GPU. The developers also created a small app to test and demonstrate these features in user space so when you're using your system logged in as a user. It's still very, very early days. This driver still cannot handle any real-world GPU at all. I don't expect it to be capable of really being usable before the end of the year. I don't expect it to become the default before the end of next year for the GPUs it handles. But I guess Red Hat has the opportunity to prove me wrong and also to prove that using Rust is actually a lot faster than using any other language to produce a fully working thing. I guess it's a very good test of Rust code in the kernel to handle a use case that is currently really not that well served. If they can manage it in a secure way with good, well-written code that is written fast and works well, then I guess they will have proven everyone wrong about Rust and people will just Have to shut up when someone says they want to write something with Rust. That would be nice to not have to talk about the language itself. Anyway, this will conclude today's episode. I hope you enjoyed listening to it. As always, if you want to dive deeper into any of these topics, you know what to do. All the links are in the show notes. If you want to support the show, there are links to do just that in the show notes as well, including a link to our sponsor, Tuxedo Computers. So thank you all for listening, and I guess you'll hear me in the next one next week. ([Time 0:27:05](https://share.snipd.com/snip/74afbb6b-ff4d-4603-876a-4e05e71efc45))