Linus Torvalds, the creator of Linux, has turned to AI-assisted coding for a hobby project, marking a shift from his earlier criticisms of such tools. In January 2026, he updated his GitHub repository AudioNoise, crediting Google's Antigravity for generating Python code to visualize audio samples. This move highlights AI's role in experimental development while he focuses on core logic in C.
In early January 2026, Linus Torvalds quietly updated his open-source project AudioNoise on GitHub, revealing his use of AI for part of the development. The project, an experimental GPL-licensed repository, explores digital audio effects such as delays and IIR filters, stemming from Torvalds' hobby of building DIY guitar effects pedals using chips like RP2354 and TAC5112. Initially focused on hardware, Torvalds shifted to digital simulation, noting in the repository that the hardware interfaces were unsatisfactory.
Torvalds wrote the core logic in C himself but delegated the Python visualization tool to AI. In the project's README, he explained: "The Python visualization tool was basically created through Vibe Coding... I skipped the intermediate step - myself - and used Google Antigravity to implement the audio sample visualization function." He described his Python knowledge as limited compared to C and analog filters, contrasting his past reliance on searching and copying code examples.
Vibe Coding, a term popularized by former OpenAI executive Andrej Karpathy, involves describing desired functions in natural language and letting AI generate the code, ideal for rapid prototyping. Torvalds employed Google's Antigravity, an experimental AI development environment released at the end of 2025, to produce the tool efficiently.
This development contrasts with Torvalds' previous stance. In 2023, amid the ChatGPT boom, he dismissed AI-generated code as lacking underlying logic and full of risks, calling reliant programmers "CV engineers." In 2024, he rejected AI for Linux kernel development, emphasizing the need for explainable code in critical systems with tens of millions of lines. However, for this non-critical hobby project, he found AI valuable, suggesting it suits experimentation but not production maintenance.
The update sparked discussion in developer communities, with observers noting it as a pragmatic evolution. Torvalds' endorsement underscores AI's maturation, from toys in 2022 to productivity tools by 2025 with models like GPT-5.2 and Gemini 3, redefining programming toward architecture and verification over line-by-line writing.
