Now FSF remembered that we're on AI hype peak: https://www.fsf.org/news/fsf-is-working-on-freedom-in-machine-learning-applications

Though I doubt much can be done by FSF in this field: even not every transnational corporation has enough resources to follow the movement, especially not to steer one.

I'd expect get some interesting initiatives in "small home easy-to-use models" like in embedded LLMs or something. Because there are some competitive models (like Llama) published to open source, but you need a rather expensive piece of hardware even for inference not to speak about tuning.

So waiting for a tiny model "free as a free mind" :)