OSes<\/p>\n
Furores are fermenting in the forums<\/p>\n
Both Ubuntu and Fedora have made it official: support is coming soon
\nfor running local generative AI instances.<\/p>\n
An epic and still-growing thread in the Fedora forums states one of Fedora Project Lead Jef Spaleta, who took I have zero evidence in front of me that users are being driven away As far as Red Hat\u2019s community distribution goes, while this may be Since Red Hat has other offerings for slow-moving stable server OSes to build a thriving community around AI technologies by focusing on He also spells out what it doesn\u2019t want to do:<\/p>\n Non-goals:<\/p>\n The system image will not be pre-configured with applications that Tools and applications included in the AI Desktop will not be AI tools will not be added to Fedora\u2019s existing system images, In other words, tools for developers, not for end-users, with a strong Our impression is that the Fedora team feels that it needs to keep Fedora is not alone in this, though. There are also ructions across the border in Ubuntuland. Right after the release of the Canonical\u2019s new LTS version,\u00a0Ubuntu 26.04 Resolute Raccoon<\/a>, Canonical\u2019s veep of engineering Jon Seager laid out\u00a0the future of AI in Ubuntu<\/a>.\u00a0<\/p>\n We\u00a0interviewed Like Fedora, there\u2019s a strong focus on local models and confidential, We are not setting shallow metrics on token usage, or percentages of Initially, the focus is on users instead:<\/p>\n AI features in Ubuntu features will come in two forms: first as a As Fernando Marcela\u2019s exit shows, an emphasis on what could be termed Back in January, we reported on the rise, Resistance, though, is also rising. Stop Slopware<\/a> tries to help explain Bootnote<\/p>\n
\nthe goals for the next version: the Fedora AI
\nDeveloper Desktop Objective<\/a>. It is causing some discontent, and at
\nleast one Fedora contributor, SUSE\u2019s
\nFernando Mancera<\/a>,\u00a0has resigned<\/a>.<\/p>\n
\nover the role<\/a> from Matthew Miller a year ago, remains resolute, saying<\/a>:<\/p>\n
\nfrom Fedora because of AI.<\/p>\n
\ncontroversial, this should not be a big shock. In October last year,
\nThe Register reported that the Fedora
\ncouncil approved a policy allowing AI-assisted contributions<\/a>, and
\nanyone following the IBM subsidiary\u2019s movements will already know that
\nlast June\u2019s RHEL 10 release includes access to an LLM-based online
\nhelper chatbot: we tried
\nit out when the product was released<\/a>. We also reported on the
\nmanagers of Red Hat\u2019s Global Engineering department being notably
\nkeen on the use of AI<\/a> just last month.<\/p>\n
\n\u2013 and arguably because Debian, Ubuntu, and their many derivatives have
\nthe stable-desktop-distro space nicely covered already \u2013 Fedora has a
\nstrong focus on providing a distro for developers, and Spaleta\u2019s
\nannouncement makes this clear. The goal is:<\/p>\n
\nthree key areas: equipping developers with the necessary platforms,
\nlibraries, and frameworks; ensuring users experience painless deployment
\nand usage of AI applications; and establishing a space to showcase the
\nwork being done on Fedora, connecting developers with a wider
\naudience.<\/p>\n
\ninspect or monitor how users interact with the system or otherwise place
\nuser privacy at risk.<\/p>\n
\npre-configured to connect to remote AI services.<\/p>\n
\nEditions, etc, by the AI Desktop initiative.<\/p>\n
\nemphasis on models that run locally, and which preserve\u00a0the user’s privacy. It\u2019s also worth pointing out that Fedora has had an AI-Assisted
\nContributions Policy<\/a> in place for six months, and earlier this
\nmonth, Fedora
\ncommunity architect<\/a> Justin Wheeler explained in some detail Why
\nthe Fedora AI-Assisted Contributions Policy Matters for Open
\nSource<\/a>.<\/p>\n
\nFedora relevant for growing interest in LLM-bot assisted tooling, and
\nthat it can address concerns from hardcore FOSS types by ensuring that
\nthis means local models, built according to FOSS-respecting terms,
\ndeployed in privacy-respecting ways.\u00a0<\/p>\n
\nSeager last year<\/a> during the 25.10 Ubuntu Summit, and back in January this
\nyear, he published his views on Developing
\nwith AI on Ubuntu<\/a>. Now the plans are firming up.<\/p>\n
\nprivacy-first deployments \u2013 and ensuring that the OS and the tools
\nsupport GPU acceleration from the big hardware players in that space.
\nHowever, unlike Red Hat, Canonical isn\u2019t pushing its developers towards
\nthese tools. In what we see as a veiled jab, Seager\u2019s announcement
\nsays:<\/p>\n
\ncode written with AI, but rather incentivising engineers to experiment
\nand understand where AI tools add value.<\/p>\n
\nmeans of enhancing existing OS functionality with AI models in the
\nbackground, and latterly in the form of \u201cAI native\u201d features and
\nworkflows for those who want them.<\/p>\n
\nFOSS-friendly AI \u2013 open models, privacy-centric, local execution and so
\non \u2013 is not enough to placate those who are really
\nstrongly averse to these tools<\/a>. The Reg FOSS desk counts
\nhimself firmly in this camp.<\/p>\n
\nfall, and resurrection of OpenSlopware<\/a>, a list of FOSS projects
\nwhich contain LLM-generated code, integrate LLMs, or even show the
\ntraces of the use of LLM agents. Soon, it seems inevitable that Fedora
\nand Ubuntu will both feature here.<\/p>\n
\nwhy and how to avoid it, and there\u2019s also The No-AI Software
\nDirectory<\/a> for projects that have\u00a0 explicit\u00a0LLM-free policies,
\nwhether they\u2019re FOSS or not.<\/p>\n