Comply with ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Canonical allows you to decide and select how you will use AI.
- In Ubuntu, AI is constructed into key options and optionally available AI instruments.
- Whereas Microsoft is about management, Canonical places you in cost.
In a brand new weblog submit, Jon Seager, Canonical’s VP of engineering for Ubuntu, defined how the corporate is baking AI into its Linux desktop and server expertise in Ubuntu Linux 26.04 and past. In contrast to Home windows, the place Microsoft is slapping its Copilot label on everything, Canonical cooks AI into its Linux distro on open terms: open fashions the place potential, native inference by default, and no rebranding of the distro into an AI product.
Seager defined that Canonical is “ramping up its use of AI instruments in a centered and principled method.” That method means a transparent choice for open‑weight fashions whose license phrases align with Ubuntu’s lengthy‑standing open‑supply values, coupled with open‑supply harnesses and tooling. The Canonical developer groups are inspired to undertake the instruments that make sense for them, so long as they select a single software persistently on the workforce degree.
Additionally: Built for a hostile internet: Canonical VP of Engineering on Ubuntu 26.04 LTS
He burdened that Ubuntu will not be being repositioned as an AI product, however that “considerate AI integration” will make the working system extra succesful and environment friendly for individuals who already depend on it. Internally, Canonical plans to teach its engineers on the place AI genuinely provides worth, and to keep away from crude metrics like “how a lot AI did you utilize,” focusing as an alternative on high quality, management, and reviewability of AI‑assisted work.
Default vs. specific AI
A core a part of Ubuntu’s framework is the excellence between “implicit” and “specific” AI options. Implicit AI options will run largely within the background, enhancing Linux’s present capabilities. That is the type of enchancment you will expertise as “the system simply works higher” fairly than as a brand new AI product. For instance, Ubuntu 26.04 boasts first‑class speech‑to‑textual content and textual content‑to‑speech, higher display screen studying, and different accessibility enhancements powered by native fashions.
Additionally: The new rules for AI-assisted code in the Linux kernel: What devs need to know
Specific AI options, against this, will arrive as new, decide‑in capabilities that clearly current themselves as AI‑pushed. These options might embody generative textual content instruments in productiveness workflows, agentic helpers for duties akin to file or mission administration, and devoted interfaces for interacting immediately with fashions. Seager describes this method as phased: first, quietly enhancing what Ubuntu already does, then layering on “AI‑native” workflows for customers who actively need them.
Don’t desire these AI-enabled packages? Tremendous. You do not have to make use of them. Good luck trying that with Windows 11.
Ubuntu is all about operating AI regionally. Canonical desires most Ubuntu AI options to default to on‑machine inference. This method makes these options usable offline, probably extra non-public, and fewer depending on proprietary cloud backends. It is going to additionally make them less expensive to make use of.
Additionally: Linux explores new way of authenticating developers and their code – how it works
This method dovetails with Canonical’s present work on tuned kernels, {hardware} enablement for GPUs and accelerators, and partnerships with silicon distributors. Seager described this as the inspiration for environment friendly native inference on abnormal Ubuntu installations.
Accessibility is among the first concrete targets for this AI push. Seager highlights system‑extensive speech‑to‑textual content and textual content‑to‑speech, plus richer display screen reader capabilities, not as flashy “AI add‑ons” however as core OS capabilities. Trying forward, he wrote, “What as we speak looks as if it is solely potential with entry to a frontier AI manufacturing facility will develop into considerably extra accessible within the coming months and years.”
Past particular person options, Canonical is pushing ahead an Ubuntu that may act as a safer house for AI brokers and agentic workflows. Seager says users are increasingly accustomed to working with agents and that he “loves the thought” of constructing the collected energy of Linux extra accessible by way of agent‑pushed interfaces. The purpose is a “context‑conscious OS” through which brokers can motive concerning the consumer’s atmosphere and duties whereas being constrained by Ubuntu’s present safety mannequin.
Additionally: ‘Like handing out the blueprint to a bank vault’: Why AI led one company to abandon open source
Right here, Snap, Ubuntu’s default utility container method, turns into Canonical’s method of securing AI brokers. With Snap, brokers will sandbox. This step blocks them from accessing restricted knowledge and sources. Canonical is exploring methods to combine such workflows “in a method that feels tasteful, aligned with our consumer base and respectful of our privateness and safety values,” explicitly acknowledging neighborhood anxiousness about heavy‑handed AI.
With Microsoft making AI a marquee branding time period, Seager is at pains to distinguish Ubuntu’s method. He rejects the thought of measuring Canonical workers by the amount of AI output, and he says the corporate will not be planning to “power” AI on customers or flip Ubuntu into an AI‑first product. On the identical time, he’s frank about AI’s impression on engineering work, noting that whereas Canonical doesn’t intend to switch individuals with AI, an engineer expert with AI instruments actually might outperform one who will not be.
One factor users should not expect is a universal “AI kill switch.” Seager argues that such a change can be complicated to implement, “truthfully,” on condition that some AI performance will blur into background system enhancements fairly than discrete apps. As an alternative, the emphasis is on preserving AI options constrained, auditable, and aligned with open‑supply expectations, whereas permitting Ubuntu to evolve in a world the place AI is quickly turning into a part of the baseline of contemporary computing.
Home windows AI vs. Ubuntu AI
Canonical is explicitly biasing Ubuntu towards open‑weight fashions, open‑supply harnesses, and mannequin licenses that align with lengthy‑standing free‑software program values, fairly than simply grabbing no matter performs greatest on benchmarks. Thoughts you, as Seager noticed, “entry to mannequin weights is significant, however it’s not equal to the kind of transparency the open supply neighborhood has develop into accustomed to.” He added, Canonical will select fashions based mostly on license phrases, not simply efficiency.
Microsoft’s mainstream AI push, against this, is anchored in proprietary cloud providers, akin to Copilot for Microsoft 365 and Azure OpenAI. Sure, Microsoft will allow you to use many fashions, however provided that Microsoft acts because the gatekeeper. You possibly can solely use AI in Home windows utilizing Microsoft’s guidelines, together with its pricing, insurance policies, and telemetry.
Additionally: The new rules for AI-assisted code in the Linux kernel: What every dev needs to know
Canonical’s plan for Ubuntu is to make native inference the default. Ideally, all AI‑enhanced OS options ought to run on units offline, with clearly outlined interfaces which can be used solely when an exterior service is genuinely wanted. That method plays to Linux’s strengthsakin to {hardware} tuning and GPU/accelerator enablement, whereas preserving your knowledge and workflows in your machines.
Microsoft’s technique has been “cloud first”: Copilot in Home windows and Microsoft 365 is essentially tied to cloud‑hosted fashions and knowledge processing, even when some shopper‑aspect NPUs become involved. That connection makes it simpler to roll out options at scale. Nonetheless, the method additionally centralizes knowledge and compute, will increase vendor dependence, and makes it tougher for customers to grasp or restrict the place their knowledge flows.
As Seager identified, Ubuntu splits AI into “implicit,” quietly enhancing present capabilities like speech‑to‑textual content, display screen studying, and different accessibility instruments, and “specific,” new, clearly labeled AI workflows or brokers that customers can select to undertake. This break up is all about AI making Ubuntu “meaningfully extra succesful” with out turning it into an AI‑branded product or forcing AI on customers who desire a secure Linux desktop.
Microsoft’s stance, alternatively, is all about pushing AI into the default consumer expertise. As an example, Copilot seems immediately within the Home windows shell and Microsoft 365 apps. As well as, Microsoft is exploring always‑on agents inside 365. There, agentic AI will act as an operational layer for workplace workflows. That shift is nice in the event you’ve already purchased into Microsoft. And, clearly, a number of persons are OK with that stance — more fool they from where I sit.
Additionally: The best free AI for coding – only 3 make the cut now
Nonetheless, being tied to Microsoft means you will need to work together with AI by default, not by a thought-about decide‑in. Are you good with that? Will you continue to be OK with it as AI costs surge higher?
Canonical’s AI story leans closely on utilizing Ubuntu’s present safety primitives, particularly Snap confinement, to offer AI brokers tightly scoped permissions, clear auditability, and totally different “grades” of entry from learn‑solely evaluation via to managed write entry. The thought is a “context‑conscious OS,” through which brokers will be highly effective, however they run inside clear, open‑supply sandboxes that customers and auditors can examine.
Additionally: I tried a command-line-only distro that can seriously improve your Linux skills
Microsoft’s agentic course is extra centered on integrating brokers immediately into enterprise workflows, akin to Microsoft 365 brokers that may act throughout mail, paperwork, and line‑of‑enterprise methods. That integration is nice for automation, however tougher for customers to grasp. Governance lives in coverage consoles and connectors that IT admins configure, not in a consumer‑seen, open safety mannequin that may be independently examined and forked.
Canonical positions Ubuntu as a low‑friction platform for local AI experimentation and open‑supply workflows. With Ubuntu, it is simple for builders to swap out fashions, frameworks, and instruments. This method makes it simpler for groups to prototype with native fashions, vector databases, and agent frameworks, and, crucially, to keep away from vendor lock-in through the experimentation section.
Microsoft’s energy is very large distribution and built-in tooling. However that very same integration makes it extra doubtless that early experiments develop into lengthy‑time period dependencies on Microsoft’s stack, with knowledge, workflows, and governance all tied to the identical vendor.
When you care about open fashions, native management, and the power to see and form how AI is wired into your system, Ubuntu is your pal. Microsoft’s mannequin is compelling for tightly coupled, cloud-first enterprise workflows, nevertheless it trades openness and portability for lock-in, deep integration, and comfort. I do know which mannequin I will be utilizing.
