I'm really disappointed to see #bitwarden falling into the slop hole. Can anyone recommend a #passwordmanager , ideally #selfhosted , that doesn't use LLM slop in its core product, OR in its contributing commits?
#askfedi

@violet I think I think keepass family is the standard here?

XC did also start doing AI, though

@astraluma fml, keepass was the chosen one. I thought it was safe. I hate it here.

I would make my own, but my specialty was python, which has now been tainted too. Fuck it aaaaallllll

@violet a few of us did a deep dive

cpython has claude contributions, but it looks like minimal code

which is honestly expected in a project that large

@astraluma my work is in connecting disparate systems, so deep level python skunk works isn't my forte. What are the ramifications of the taint moving forward, in your opinion?

For context, I'm wary of Zen browser for being forked from FF for having any slop taint, so these reservations aren't specific to python

@violet I think it's mostly "a couple of people were sloppy about its use"

none of the regular cpython contributors seem to be using AI in easily tracable ways. There's about a dozen suspect commits, and they're all small changes in dusty corners.

Predicting going forward is going to depend a lot on how you feel about the prospects of AI in development.

@violet My current semi-reasoned prediction is that the bubble is going to pop and LLMs are going to get dramatically more expensive to use.

As tokens become a meaningful resource, I expect less of them will be spent on FOSS work.

As someone who doesn't like AI and is quite ready for this bubble to pop and this hype cycle to end, I'm hoping this will mean that cpython remains minimally tainted.

@violet at this point, I don't think it's reasonable to do non-trivial computing without some software that's been touched by AI in some way.