John O'Nolan (@index.john.onolan.org.ap.brid.gy)

Open Source in the age of AI Early 2026 is a very weird time to be an open source maintainer. On the one hand, the burden of codebase maintenance has dropped dramatically. Small teams with long todo lists now have the ability to accomplish more than ever before. On the other hand, the dynamics of the software industry are rapidly changing, and some of the foundations open source projects built upon are looking increasingly shaky. Any open source project that achieves a degree of success will at some point face the reality of a competing company taking your open source code and using it to compete with you. Substack has – at various points – wholesale copied and pasted significant chunks of Ghost's source code into its own product. Open source maintainers have been fighting this sort of thing for a while with increasingly restrictive licenses, or keeping certain parts of the development stack private, like the test suite. "If you want to use our code, you have to follow our rules." But the dictating of the rules has largely been built upon a premise that code is intrinsically valuable because it's a scarce resource. Difficult to maintain. Expensive to write. AI is now rapidly overturning that premise, as evidenced by Cloudflare this week, who took a competitor's open source codebase and cloned it in the space of a few days into an entirely new product: How we rebuilt Next.js with AI in one weekOne engineer used AI to rebuild Next.js on Vite in a week. vinext builds up to 4x faster, produces 57% smaller bundles, and deploys to Cloudflare Workers with a single command.The Cloudflare BlogSteve Faulkner If anyone can point Claude at an open source codebase and have it rewritten from scratch, without actually using any of the original code, what does that mean for software licenses? More specifically: Do software licenses mean anything? Some will debate whether this is even a thing – with a lot of hand waving about "vibecoding slop" – but the trajectory is clear to see. In December, Simon Willison wrote about porting an HTML parsing library from Python to JavaScript, raising these legal/ethical concerns as a thought exercise. Two months later, we have a publicly traded company shamelessly cloning the entire product of a competitor purely for the sake of strategically undermining them. In some ways this feels like the start of software's Studio Ghibli moment. But it would be a mistake, I think, to imagine that only open source software will be susceptible to being cloned. I don't imagine it will be long before any sufficiently successful proprietary product with a public-facing interface can be reverse-engineered and rebuilt by a motivated competitor with access to frontier models. What took Cloudflare a week with an open source codebase will likely soon take a week to do with a closed one. LLMs, it turns out, are already pretty good at reverse engineering. the z80 technique reveals the source code for Atlassian’s ‘rovo’ AI assistantEver wondered what happens if you take the technique at “Can a LLM convert C, to ASM to specs and then to a working Z/80 Speccy tape? Yes.” and run it against the Atasslian Command Line (ACLI) interface? Strap yourself in, as the Z80 is amongst one of theGeoffrey HuntleyGeoffrey Huntley Which is an interesting wrinkle. What happens when we get to a point where closed products can just as easily be cloned and released as open source? When Stallman launched the free software movement in 1983, it was a reaction to a world where software was closed, proprietary, and wielded as an instrument of control. Open source was built upon the idea that anyone should be able to study, modify, replicate and share software. Free from the grasp of corporations and copyrights to prevent it. In a strange way, I'm beginning to wonder if AI might end up fulfilling that vision more completely than open source ever did. https://john.onolan.org/open-source-in-the-age-of-ai/