1. The Kernel Maintainer’s Official Pivot on AI
Linux creator Linus Torvalds has drawn a definitive line in the sand regarding artificial intelligence in the Linux kernel. In a recent exchange on the kernel mailing list, Torvalds made it clear that the project will not adopt an anti-AI stance, regardless of developer sentiment. He emphasized that AI is now a standard engineering tool, comparable to compilers or version control systems, and that its utility in code review is no longer debatable.
This declaration comes after a heated debate surrounding Sashiko, an agentic review tool developed by Google engineer Roman Gushchin. Sashiko analyzes submitted kernel patches and independently flags potential bugs before human reviewers ever see them. According to Gushchin, the system catches 53.6 percent of issues that would otherwise slip past experienced maintainers. When kernel developer Laurent Pinchart requested a human filter to screen Sashiko’s output, Gushchin pushed back, arguing that bypassing the tool defeats its purpose.
Torvalds sided with Gushchin, stating, “I realize that some people really dislike AI, but this is an area where I’m willing to absolutely put my foot down as the top-level maintainer. Linux is not one of those anti-AI projects.” He acknowledged that while AI adds workload by surfacing bugs that require human resolution, the alternative is falling behind in code quality. The decision underscores a broader industry shift: AI in software development is no longer experimental, it is operational.
2. From Skepticism to Adoption: The One-Year Gap
Torvalds’ current position stands in stark contrast to his remarks at the Open Source Summit in Vienna back in October 2024. At the time, he dismissed the AI wave as “90 percent marketing and 10 percent reality,” advising developers to wait out the hype cycle. That assessment was rooted in the limitations of early large language models, which frequently hallucinated code, struggled with context, and produced unreliable output for complex systems.
Over the past twelve months, however, the underlying models have undergone significant architectural improvements. Tools like Claude, ChatGPT Codex, and specialized code review agents now demonstrate a level of precision that makes them viable for production environments. The Linux kernel, known for its rigorous quality standards and strict review process, now permits AI to comment directly on patches from developers with decades of experience. Torvalds noted that the tools he once dismissed are now “clearly a useful one,” a concession that reflects the rapid pace of AI advancement rather than a change in his engineering philosophy.
3. Microsoft’s AI-Code Admission and the Windows 11 Backlash
While the open-source community debated AI’s role in kernel development, Microsoft was quietly integrating AI into its own engineering pipeline. In April 2025, CEO Satya Nadella revealed at Meta’s LlamaCon that 20 to 30 percent of Microsoft’s codebase was already AI-generated. Google’s Sundar Pichai made a nearly identical disclosure that same month, signaling a cross-industry pivot toward AI-assisted development.
The announcement triggered immediate backlash. Critics pointed to Microsoft’s history of overpromising on emerging technologies, particularly during Windows 11’s notoriously difficult 2025. That year saw more than twenty major update failures, ranging from USB audio and webcam regressions in January to BitLocker settings failing under security protocols, and a mandatory March update that triggered BSODs and boot loops. Remote Desktop sessions dropped mid-call after a February patch worsened the March release, while Task Manager began spawning duplicate processes that refused to terminate.
By January 2026, Microsoft publicly acknowledged that Windows 11 had “gone off track,” attributing the instability to a focus on Agentic OS ambitions over foundational stability. A former Microsoft VP, drawing on experience at Google DeepMind and Meta, later noted that Copilot’s paid adoption sat at a mere 3.3 percent across 450 million Microsoft 365 users, suggesting the company had missed the AI wave just as it had missed the internet and mobile eras. The correlation between heavy AI integration and Windows 11’s worst performance year was impossible to ignore, even if Microsoft stopped short of blaming AI code for specific bugs.
4. The 2026 Turnaround: AI as a Bug-Hunting Weapon
Windows 11’s recovery began in earnest during the spring of 2026. Windows president Pavan Davuluri published a roadmap on March 20 outlining concrete fixes for File Explorer, the Widgets board, and Copilot integration. By May, testers reported fewer forced restarts, extended update pause options, and a File Explorer that no longer flashed white when toggling between light and dark modes. The June 2026 Patch Tuesday delivered the largest update of the year, adding CPU performance boosts and long-requested Bluetooth stability fixes.
The most significant development, however, came in July 2026. Microsoft patched a record 570 security flaws in a single update, a 316 percent increase from July 2025. The company credited MDASH, a new AI scanning system that leverages a combination of models including Anthropic’s, for identifying vulnerabilities before attackers could exploit them. Microsoft’s explanation highlighted how AI is compressing the window between vulnerability discovery and exploitation, shifting the defense model from reactive to predictive.
This improvement mirrors the same AI advancements that convinced Torvalds to embrace kernel review tools. Both projects are now using AI not to replace human judgment, but to amplify it. Where 2025 saw AI as a liability, 2026 is proving it to be a critical component of modern software maintenance.
5. What This Means for You
For everyday Windows 11 users, the shift toward AI-driven development has tangible implications. The record patch volume in July 2026 suggests that security updates will become more frequent and comprehensive, reducing the window of exposure to zero-day exploits. However, it also means that Windows Update will continue to evolve, requiring users to stay current with Microsoft’s new release cadence.
If you’re running Windows 11, ensure that automatic updates are enabled and that you’re not manually pausing them for extended periods. The improvements in stability, File Explorer performance, and Bluetooth reliability are rolling out through standard Patch Tuesday releases, so keeping your system current is the fastest way to benefit from these changes. For enterprise administrators, the MDASH system will likely influence future compliance and vulnerability management strategies, as AI-scanned patches may require updated testing protocols before deployment across fleets.
6. The Bigger Picture: AI as a Tool, Not a Replacement
Projects like Curl and Gentoo have restricted AI contributions due to quality concerns, but those restrictions stem from past failures, not a rejection of the technology itself. Torvalds’ stance reflects a pragmatic engineering reality: AI only earns its place if it makes the human developer’s job easier. In 2025, that balance was off. In 2026, it has finally tipped in favor of adoption.
Microsoft’s journey from AI-driven instability to AI-assisted stability offers a cautionary tale for the broader industry. Early adoption without proper testing can backfire, but sustained investment in AI tools, when paired with rigorous human oversight, yields measurable results. As both the Linux kernel and Windows 11 continue to integrate AI into their development pipelines, the focus will shift from whether AI should be used to how it can be used more responsibly. The tools are here, they are improving, and the developers who refuse to adapt will find themselves falling behind.
Source: Windows Latest
Over to you: Are you seeing noticeable stability improvements in Windows 11 after the July 2026 patch, or are you still running into issues?
You may also like
Gears of War: E-Day partners exclusively with NVIDIA for PC graphics
New Outlook warns against replying to old emails, adds faster categorization
Microsoft retains #2 spot in Interbrand’s Best Global Brands despite Xbox layoffs
id Software was developing a John Wick-style shooter and Perfect Dark before Xbox layoffs
