)
Perspectives
How Social Income Contributed to ChatGPT
)
Lea StrohmLoading Social Income...
Our work
Recipient app
© 2026 Social Income · Registered Non-Profit in Switzerland
Platform partner
){kind=logo}
© 2026 Social Income · Registered Non-Profit in Switzerland
Loading Social Income...
Loading Social Income...
© 2026 Social Income · Registered Non-Profit in Switzerland
Platform partner
© 2026 Social Income · Registered Non-Profit in Switzerland
Loading Social Income...
Lea StrohmÜbersetzt aus demEnglischen
That openness is also what the big AI companies lean on, some would say abused, to build their billion dollar language and coding models.
Because Social Income’s codebase is also hosted publicly on GitHub, we can be reasonably sure it was used to train Large Language Models (LLMs) such as ChatGPT and Claude. And we can even see it in our data:
“6.79k requests from AI Crawlers”
It is a strange loop: a platform designed to redistribute wealth became a small brick in the foundation of companies now worth hundreds of billions of dollars.
And now the very companies that were built on open source are warning (or boasting, depending on who is speaking) that AI will lead to large-scale job displacement. That prospect has prompted a growing number of economists to call for universal basic income as a response. The very thing Social Income exists to deliver. We did not plan for this moment. The technology created it.
But AI did not just borrow from open source, it also changed what open source feels like from the inside, for better and worse.
The good part is real. Coding is faster, including for us. Features that used to take a volunteer a weekend now take an evening. That matters when your entire engineering team has day jobs: It is the difference between shipping something this month or shelving it until someone finds the time.
Take our new website for example, which launched in the summer of 2026: Some of the best animations — like the globe showing where donations came from — simply wouldn’t exist without AI.
){kind=logo}
von 40+ open source developers (MIT License)
The strange part is also real. Open source communities used to be groups of humans writing code for other humans. Now a contribution might come from someone who is not made of flesh at all, just a language model doing the work behind a friendly username.
Or it might come from an actual human who is mostly forwarding whatever the model generated. The two are not always easy to tell apart.
That puts pressure on the humans left holding the project together. AI contributions are often genuinely good — which is precisely what makes them hard to refuse. But goodwill has limits, and so does volunteer time.
Wave things through carelessly and eventually something dangerous lands in code that moves real money. A rounding error. A silent failure in a transfer. The kind of bug that passes every test and surfaces only when someone gets hurt.
There is no clean answer yet. We are figuring it out in public, like everyone else. As a first step, we added Claude.md files to our repository, so AI tools better understand the project’s context and structure before they start proposing changes.
If you are a developer interested in doing something useful with your time, the codebase is on GitHub. We still need humans — people who read the code, understand what it does, and take responsibility for what they ship. Whether or not you use AI to help you get there.
Get to know great open source developers of Social Income: Open Source Community
This article was updated on Sept 13, 2026 with data from Cloudflare, our web analytics and CDN provider.
Lea Strohm)
)
:format(webp))
)
)