Listen to this Post
Introduction: The AI Debate Has Entered a New Era
Artificial intelligence has become one of the most polarizing technologies in modern software development. While some developers view it as the future of programming, others fear it will compromise software quality, security, and the principles of open-source development. Few voices carry as much influence in this debate as Linux creator Linus Torvalds, whose opinions often shape discussions across the global developer community.
For years, Torvalds remained skeptical of AI, dismissing much of the industry’s excitement as little more than marketing hype. However, recent events reveal a remarkable shift in his perspective. Rather than rejecting AI, Torvalds now openly supports its use within Linux kernel development—as long as it proves its technical value.
Ironically, this change comes at a time when Microsoft, one of AI’s biggest corporate champions, is finally beginning to demonstrate that AI can improve software quality after enduring one of Windows 11’s most turbulent periods. Together, these developments suggest that the conversation surrounding AI has evolved from “Should we use AI?” to “How should we use AI responsibly?”
Linus Torvalds Declares Linux Is Not an Anti-AI Project
The controversy began after discussions surrounding Sashiko, an AI-powered patch review assistant developed in part by Google engineer Roman Gushchin. The tool analyzes Linux kernel patches and automatically detects bugs before human maintainers approve code.
According to its developers, Sashiko successfully identifies more than half of the issues that would otherwise escape traditional review until much later in the development cycle.
Not everyone welcomed the technology.
Several Linux developers argued that AI-generated comments should first be filtered by experienced humans before reaching contributors. Their concerns echoed recent recommendations from open-source organizations warning against blindly trusting AI-generated output.
Torvalds disagreed.
He made it clear that Linux would not become an anti-AI project simply because some contributors dislike large language models.
Instead, he emphasized that AI should be judged exactly like any other engineering tool—by its technical performance rather than emotional reactions.
For Torvalds, software decisions must always prioritize engineering merit over ideology.
A Dramatic Change From His Earlier Position
Only eighteen months earlier, Torvalds had described artificial intelligence as being “90 percent marketing and 10 percent reality.”
At the time, his criticism reflected the actual capabilities of AI coding assistants. Many tools generated inconsistent code, hallucinated APIs, and frequently introduced security issues.
Fast forward to today, and the landscape looks dramatically different.
Modern coding assistants such as ChatGPT Codex, Claude, Gemini, and other advanced models have significantly improved their reasoning, debugging, code review, and vulnerability detection capabilities.
The fact that Linux kernel maintainers—known for maintaining one of the strictest software review processes in the world—are now allowing AI to review kernel patches demonstrates just how much these technologies have matured.
When someone as notoriously difficult to impress as Linus Torvalds changes his opinion, the software industry pays attention.
Microsoft Experienced the AI Revolution Much Earlier
Long before Linux officially embraced AI-assisted development, Microsoft had already integrated artificial intelligence deeply into its engineering workflow.
In 2025, Microsoft CEO Satya Nadella revealed that approximately 20–30% of Microsoft’s code was already being generated with AI assistance.
Google CEO Sundar Pichai made similar statements regarding Google’s internal software development.
Instead of admiration, however, Microsoft faced heavy criticism.
Many users assumed Windows itself was becoming AI-written, creating fears that quality would inevitably decline.
Microsoft later clarified that Windows was not being entirely rewritten by AI, although developers were increasingly relying on AI-assisted coding tools.
Windows
Unfortunately for Microsoft, the timing could not have been worse.
Throughout 2025, Windows 11 suffered one of its most unstable periods since launch.
Major updates repeatedly introduced serious problems, including:
USB device failures
Webcam compatibility issues
BitLocker configuration bugs
Blue Screens of Death
Boot loops
Remote Desktop instability
File Explorer glitches
Task Manager duplication bugs
These issues created the public perception that AI-generated code was responsible.
However, Microsoft never presented evidence connecting AI-assisted programming directly to those failures.
The widespread assumption remained speculative rather than factual.
Still, perception often matters as much as reality.
Microsoft Begins Recovering Through Better Engineering
By early 2026, Microsoft publicly acknowledged that Windows development priorities had drifted too far toward ambitious AI initiatives while neglecting core operating system stability.
Windows leadership responded with a roadmap focused on fixing everyday usability problems instead of introducing flashy AI features.
Subsequent updates delivered meaningful improvements.
Microsoft optimized File Explorer performance, reduced unnecessary Copilot integrations, improved update controls, enhanced Bluetooth reliability, increased CPU performance, and reduced forced system restarts.
Most importantly, Windows updates became noticeably more stable.
Rather than marketing AI as the headline feature, Microsoft quietly used AI internally to improve software quality.
AI Is Now Helping Microsoft Find Security Bugs Faster
One of
The company fixed 570 security vulnerabilities, the largest Patch Tuesday in Microsoft’s history.
Microsoft credited much of this success to MDASH, an AI-assisted vulnerability discovery platform combining multiple machine learning models.
Instead of replacing security researchers, MDASH accelerates vulnerability discovery before attackers can exploit software weaknesses.
This represents perhaps the strongest real-world example of AI serving as a force multiplier rather than a replacement for experienced engineers.
Ironically, this practical success mirrors exactly the reasoning behind Torvalds’ acceptance of AI in Linux development.
Linux and Windows Are Quietly Following the Same Philosophy
Although Linux and Windows often represent opposing ecosystems, both projects are arriving at the same conclusion.
Artificial intelligence is not replacing software engineers.
It is becoming another engineering tool.
Like static analyzers, automated testing frameworks, fuzzing systems, and compilers, AI earns its place only when it consistently improves software quality.
Projects such as Curl and Gentoo continue restricting AI-generated submissions—not because AI itself is unacceptable, but because every contribution must still satisfy rigorous quality standards.
That philosophy aligns perfectly with
Human expertise remains the final authority.
AI simply helps humans work faster and catch mistakes earlier.
The Industry Is Moving Beyond the AI Hype Cycle
The software industry appears to be leaving behind the exaggerated promises that dominated AI discussions throughout 2024 and much of 2025.
Developers no longer judge AI based on marketing presentations.
Instead, they evaluate measurable outcomes.
Can AI detect vulnerabilities?
Can it review pull requests accurately?
Can it improve productivity without introducing additional bugs?
Can it reduce repetitive engineering work?
These are the questions now shaping enterprise adoption.
The debate has matured considerably.
Deep Analysis
The evolution of AI-assisted software engineering reveals an important distinction between AI-generated code and AI-assisted development. Mature software organizations increasingly use AI as an intelligent reviewer rather than an autonomous programmer.
Typical secure development workflow:
Developer Writes Code
│
▼
AI Static Analysis
│
▼
Human Code Review
│
▼
Automated Testing
│
▼
Security Scanning
│
▼
Continuous Integration
│
▼
Production Deployment
Useful security and development commands commonly used alongside AI-assisted workflows include:
git diff git log git blame git bisect make test clang --analyze cppcheck . scan-build make cargo audit npm audit pip-audit
go test ./…
Kernel developers also combine AI recommendations with automated validation systems before merging changes:
git format-patch git apply git am make oldconfig make -j$(nproc) make modules_install make install
AI should never bypass these verification stages. Instead, it complements existing quality assurance processes.
The Linux kernel demonstrates one of the
Microsoft’s security improvements also illustrate another trend: AI excels at identifying repetitive patterns across millions of lines of code. This allows security researchers to focus on complex logic flaws rather than manually searching for common vulnerabilities.
The lesson is becoming increasingly clear.
AI performs best when humans remain responsible for design decisions, architecture, validation, and final approval.
The future of software development is unlikely to be human versus AI.
Instead, it will be human with AI.
What Undercode Say:
The most interesting part of this story is not that Linus Torvalds changed his opinion.
The real story is why he changed it.
Torvalds has always valued measurable engineering results over trends.
He ignored AI when it mostly generated impressive demonstrations but inconsistent production code.
Now he sees practical improvements backed by evidence.
That is a fundamentally different situation.
Microsoft experienced the opposite timeline.
The company aggressively promoted AI before its practical benefits became obvious to users.
This created unrealistic expectations.
When Windows 11 encountered stability problems, many people immediately blamed AI despite lacking evidence.
The coincidence damaged public trust.
Today the situation looks different.
Modern language models perform far better at code reasoning, vulnerability discovery, documentation, and debugging.
Large software companies are no longer asking whether AI works.
They are asking where AI delivers measurable value.
Linux and Microsoft now appear surprisingly aligned.
Neither wants AI making final decisions.
Neither wants AI replacing experienced developers.
Both want AI handling repetitive analysis while humans retain authority.
This hybrid approach is likely to dominate enterprise software development over the next decade.
Another overlooked point is developer productivity.
AI reduces time spent reviewing boilerplate code, searching documentation, identifying syntax errors, and detecting common security flaws.
That allows engineers to focus on architectural decisions requiring human creativity.
Security also benefits.
Attackers already use AI to discover vulnerabilities.
Defenders must respond with equally capable defensive AI.
Otherwise, cybercriminals gain a growing advantage.
This creates an AI arms race in cybersecurity.
Organizations refusing AI entirely may eventually struggle to compete against attackers using automated vulnerability discovery.
However, blind trust in AI remains dangerous.
Hallucinated code, incorrect recommendations, and subtle security mistakes still occur.
Every AI-generated suggestion requires validation.
Responsible adoption—not blind adoption—is becoming the
Ultimately,
That philosophy has always defined Linux.
Now it also defines its approach toward artificial intelligence.
✅ Fact: Linus Torvalds publicly stated that Linux is not an anti-AI project and defended the use of AI tools when they demonstrate technical value. This aligns with his recent comments regarding AI-assisted kernel development.
✅ Fact: Microsoft previously disclosed that AI assists in writing a portion of its internal codebase. Similar disclosures have also been made by Google regarding AI-assisted software development.
❌ Unproven Claim: There is no verified evidence that AI-generated code directly caused Windows 11’s widespread bugs during 2025. While many users associated the timing of Microsoft’s AI announcements with Windows instability, Microsoft has never confirmed a causal relationship. Correlation should not be interpreted as proof of causation.
Prediction
(+1) AI-assisted software development will become a standard component of major open-source and commercial projects within the next few years, especially for code review, vulnerability detection, automated testing, and documentation generation.
(-1) Organizations that overestimate
▶️ Related Video (72% Match):
🕵️📝Let’s dive deep and fact‑check.
🎓 Live Courses & Certifications:
Join Undercode Academy for Verified Certifications
🚀 Request a Custom Project:
Secure, high-velocity infrastructure and disruptive technological engineering. Contact our engineering team for high-tier development and proprietary systems:
[email protected]
💎 Smart Architecture | 🛡️ Secure by Design | ⭐ Trusted by Thousands
References:
Reported By: www.windowslatest.com
Extra Source Hub (Possible Sources for article):
https://www.quora.com
Wikipedia
OpenAi & Undercode AI
Image Source:
Unsplash
Undercode AI DI v2
🔐JOIN OUR CYBER WORLD [ CVE News • HackMonitor • UndercodeNews ]
📢 Follow UndercodeNews & Stay Tuned:
𝕏 formerly Twitter 🐦 | @ Threads | 🔗 Linkedin | 🦋BlueSky | 🐘Mastodon | 📺Youtube
