Fuck. I got a reply from 1password support regarding my question if they use AI/LLM for developing the password manager. The answer was affirmative, they use it for writing code, documentation och code reviews.
They claim that it's not AI-driven in the sense that it's replacing the humans but supporting them.
While they did not specify what models they use in the e-mail, a blog post they wrote quotes a guy talking about frontier models through cursor so I'll just assume they are using them rather than local models... fuck.