As expected, having Brian with us shaped most of the discussion around the past, present, and future of Open Source in the AI era. When you put a founder of The Apache Software Foundation and board member of the Electronic Frontier Foundation (EFF) and Mozilla Foundation together with engineers, researchers, and founders, some of whom are building open data language models like pleias, contributing to open source projects, or working in health regulated industries, the passionate conversation naturally goes far beyond benchmarks.
Some of the topics were digital sovereignty and the “kill switch” risks tied to the US, highlighting why “freedom of choice” and optionality are now mandatory risk management for European companies. We also had a discussion on the “Joyless Revolution” of AI, where Brian compared the current atmosphere to the Internet and Open Source revolutions that were fueled by the joy of connection and the pursuit of liberty. In contrast, today’s AI headlines are often dominated by cynicism and corporate layoffs, highlighting the urgent need to invent use cases that actually serve social good and sustainability.
Of course, we discussed vibe coding and PR Slop that is getting out of hand: the struggle of open source maintainers (like those at curl and VLC) being hammered by low quality, AI generated contributions. As Brian joked, we may soon see every senior engineer becoming a “Vibe Coding Cleanup Specialist.”
We also talked about why the future of decentralized AI might live in small reasoning models (below 1b parameters) that can run on a Raspberry Pi without a cloud connection. Anastasia warned that companies need to be careful with proprietary fine tuning on top of closed licenses. It can be a trap for the enterprise, or “paying for the wind” as she said.