The actual features they list don’t seem bad. Better text to speech and speech to text are good. Making it possible to install AI models via Snap is whatever (who the fuck uses Snap anyway). Making it easier for AI “agents” to read the system config is… well it doesn’t impact anyone not using AI, but it’s useful to people doing something deeply irresponsible in the first place, so I can’t say it’s neutral.

I don’t really think it’ll impact the wider ecosystem much. It’s actually kind of good to have Canonical wasting their time on this shit, instead of trying to introduce new standards that everyone has to scramble around and catch up with, again.

I’ve previously described Canonical as “not evil, just your cat that pukes on the carpet sometimes,” and we might have to upgrade them to “your uncle who’s handy, but starts saying some sus things when he’s drunk, and he’s been drinking a lot lately.” If you’re super involved in OSS then maybe it’s time to dust off LMDE and file bugs or whatever.