One of the things Jay laid out at SXSW earlier this week. Bluesky has a robots.txt-like proposal to allow users themselves to declare if they're okay if they're data is used for AI training. - ThreadSky

mmasnick.bsky.social • 20 hours ago

One of the things Jay laid out at SXSW earlier this week. Bluesky has a robots.txt-like proposal to allow users themselves to declare if they're okay if they're data is used for AI training.

Reposted from Jay 🦋

We put up a proposal that lays out a way for users to declare whether/how they want their data to be used by things like generative AI or public archives, check it out on github:

github.com/bluesky-soci...

Comments

zencker.bsky.social•16 hours ago

I just wish I could get paid for it…

sagcap1222.bsky.social•20 hours ago

No one trusts companies to respect their "Disallow" line in their profile!

"Companies and research teams building AI training sets are expected to respect this intent when they see it, either when scraping websites, or doing bulk transfers using the protocol itself."

mmasnick.bsky.social•19 hours ago

It has worked with robots.txt. Those who don't respect it tend to be ostracized and it's been pretty effective.

darkoverlordofdata.com•20 hours ago

in this case:
"Realistically, a large majority of users may stick with the default "undeclared" state."
it doesn't really do anything I would like to see the default to be "disallow" so that accounts have to opt in this opt "undefined" is lame

mmasnick.bsky.social•20 hours ago

Bah. "their." Github proposal for edits when?

loque.bsky.social•8 hours ago

I do wonder how much these things matter - from my perspective if something is public, it can be scraped (regardless of legality).

cdpositive.bsky.social•19 hours ago

We had *edit* on Post. No idea how difficult or cumbersome it was to implement, but there was a time limit to make changes, then the post was frozen. That said, I'm certainly not leaving just because it's not here.... As for AI training? No, thanks....

mmasnick.bsky.social•15 hours ago

There are a bunch of tradeoffs to consider which make it a bit more difficult to implement than people think. But, yeah, it can be done, and I believe will be done.

swetland.frotz.net•20 hours ago

Edit would be nice. It's 2025. Edit is a solved problem. Except for here for some reason.

russ.amodernremedy.com•17 hours ago

Edits & Polls, Edits & Polls.

thecleric.bsky.social•20 hours ago

I propose we can edit tweets via git commits.

williamcb.bsky.social•19 hours ago

Opt in or opt out?

timsewell.bsky.social•19 hours ago

You can edit, but only via pull request.

mathiastck.bsky.social•18 hours ago

LGTM

onedimental.bsky.social•17 hours ago

"If their ok" ?

That still isn't right! I thought you worded words for a living!

multichemex.bsky.social•20 hours ago

I have hope for this platform ✊🏽🤙🏽

mrbowlman.bsky.social•20 hours ago

It’s ironic that I’d be more ok with someone who respects a robots.txt scheme using my data to train ai than someone who doesn’t, but the mechanism has the opposite result.

Comments

Posting Rules

Reply