Software developers have created a PowerShell script to remove AI features from Windows.
The script, available at a GitHub account belonging to a developer named "zoicware," is called Remove Windows AI.
"The current 25H2 build of Windows 11 and future builds will include increasingly more AI features and components," the project's GitHub repo explains. "This script aims to remove ALL of these features to improve user experience, privacy and security."
The script, which its developers have worked on since 2024, includes contributions from devs "adeel26in" and "csmit195." The repo warns that third-party antivirus software "will falsely detect the script as malicious" and advises those unsure about the code to test it on a virtual machine.
The Register believes only individuals capable of reviewing the code should consider running this sort of script.
This effort is like projects such as Win11Debloat, and started attracting attention last month. It was celebrated recently on social media service BlueSky by Meredith Whittaker, president of secure messaging service Signal.
Whittaker described it as a "community-created harm reduction infrastructure to contest the alarming integration of AI agents into the Windows operating system (which is currently the most reckless deployment environment)."
Signal's president has spoken previously about the threat that AI-enabled systems – particularly agentic AI – pose to application security and privacy. Addressing the 39th Chaos Communication Congress in December 2025, Whittaker urged people to stop the reckless deployment of AI features like Microsoft Recall that violate security and privacy norms.
"We're seeing plain text databases accessible to malware, insecure storage that ignores principles of least privilege, screen recording features like those we had to jankily defend against with Recall, and the creation and aggregation of new and invasive forensic data and other personal data that is putting us all at risk," she said.
Security and privacy are not the only issues people raise about AI. Some object to the legality and ethics of using models trained on data scraped without consent, or to the market impact of having one's creative work captured, laundered, and sold back in the form of an API. Others object to AI datacenters’ environmental impact, in terms of water and electricity consumption and CO2 emissions.
The Register has also reported concerns that lax regulation of AI, and its growing capabilities, can create existential threats. Another vein of discontent addresses the bias and errors arising from AI models, and the lack of public accountability for distributing AI-generated sexual abuse material. Some worry that reliance on AI capabilities will cause critical thinking skills to atrophy or never develop at all.
And then there’s the argument that AI-generated content muddies the waters of public discourse with slop and helps to spread misinformation.
That said, many professional software developers have come to appreciate AI assistance for certain coding tasks. And companies are finding ways to use AI in public-facing applications.
Much of the animus against AI is directed at Microsoft because it controls the dominant commercial desktop operating system, has a history of annoying Windows customers by allowing the pre-installation of unwanted bloatware, and has compounded that reputation by injecting AI into as many applications as possible with few concessions to critics.
Microsoft's insistence on implementing AI features despite pushback has prompted ongoing objections, in the form of public posts and petitions.
Microsoft CEO Satya Nadella at the start of the new year published a blog titled "sn scratchpad" in which he urged people to move past arguments about AI slop and focus on how AI models can amplify human activity. He hasn't posted anything since, but there's no sign his single exhortation to dial-back the AI disparagement has swayed detractors’ opinions.
Apple, the other major commercial desktop operating system vendor, stumbled out of the AI gate and now finds itself playing catch-up rather than battling against backlash.
But other AI companies are also keen to convince customers to use their AI services. They've invested huge amounts of money building out data centers to accommodate anticipated demand and if they can't attract enough paying customers on AI product merits, they'll likely fall back on AI product myths, such as the notion that AI will enhance productivity.
AI services can, in specific scenarios, make workers more productive, but on the whole, they create as many problems as they solve. As noted in a July 2025 review of 37 studies looking at large-language-model assistants for software development, "While AI can accelerate individual work, meta-analytic evidence finds no robust relationship between AI adoption and aggregate productivity gains."
Microsoft meanwhile would very much like to find a relationship between AI investment and increasing growth. ®
Source: The register