[bhouser]

That’s my understanding as well, that it’s opt-in. That brings me little comfort.

I actually love the idea of embedded LLM help. I just wish Google used on-device compute with small models.

The idea that the browser will be used to farm data that’s not on the public internet is reprehensible to me, consent notwithstanding.

When I hire a junior engineer do I have to teach them about Google’s dark patterns so they don’t unwittingly upload our intellectual property?