\n<\/p>\n
Social network Bluesky recently published a proposal on GitHub<\/a> outlining new options it could give users to indicate whether they want their posts and data to be scraped for things like generative AI training and public archiving.<\/p>\n CEO Jay Graber discussed the proposal earlier this week,<\/a> while on-stage at South by Southwest, but it attracted fresh attention on Friday night, after she posted about it on Bluesky<\/a>. Some users reacted with alarm to the company\u2019s plans, which they saw as a reversal of Bluesky\u2019s previous insistence that it won\u2019t sell user data to advertisers<\/a> and won\u2019t train AI on user posts<\/a>.<\/p>\n \u201cOh, hell no!\u201d the user Sketchette wrote<\/a>. \u201cThe beauty of this platform was the NOT sharing of information. Especially gen AI. Don\u2019t you cave now.\u201d<\/p>\n Graber replied<\/a> that generative AI companies are \u201calready scraping public data from across the web,\u201d including from Bluesky, since \u201ceverything on Bluesky is public like a website is public.\u201d So she said Bluesky is trying to create a \u201cnew standard\u201d to govern that scraping, similar to the robots.txt<\/a> file that websites use to communicate their permissions to web crawlers.<\/p>\n Debates about AI training and copyright have dragged robots.txt into the spotlight<\/a>, among other things highlighting the fact that it\u2019s not legally enforceable. Bluesky frames its proposed standard as one that would have a similar \u201cmechanism and expectations,\u201d providing \u201ca machine-readable format, which good actors are expected to abide, and does carry ethical weight, but is not legally enforceable.\u201d<\/p>\n Under the proposal, users of the Bluesky app, or other apps that use the underlying ATProtocol<\/a>, could go into their settings and allow or disallow the usage of their Bluesky data across four categories: generative AI, protocol bridging (i.e., connecting different social ecosystems), bulk datasets, and web archiving (such as the Internet Archive\u2019s Wayback Machine).<\/p>\n If a user indicates that they don\u2019t want their data used to train generative AI, the proposal says, \u201cCompanies 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.\u201d<\/p>\n Molly White, who writes the Citation Needed newsletter and Web3 is Going Just Great blog, described this<\/a> as \u201ca good proposal,\u201d and said it was \u201cweird to see people flaming BlueSky for it,\u201d since it\u2019s not so much \u201cwelcoming in AI scraping\u201d but rather \u201ctrying to add a consent signal to allow users to communicate preferences for the scraping that is already happening.\u201d<\/p>\n \u201cI think the weakness with this and [Creative Commons\u2019] similar proposal for \u2018preference signals\u2019 is that they rely on scrapers to respect these signals out of some desire to be good actors,\u201d White continued. \u201cWe\u2019ve already seen some of these companies blow right past robots.txt or pirate material to scrape.\u201d<\/p>\n<\/div>\n