here's the best writeups we have so far.

matt's meltdown:

tumblr's AI deals:

that second article requires an account, but the full article is on tumblr here.

importantly, the initial data dump for OpenAI included a bunch of content that — on top of all public content between 2014 and 2023 — automattic didn't mean to share, like private posts, posts on suspended or deleted blogs, unanswered or private asks, and posts that are marked NSFW.

automattic got the IDs for these posts after providing them, and requested they be removed. we don't know for sure that they have been.

poking a hornet's nest here but answering a few questions i've been posed.

part of what makes this bad is that automattic already sent the data to at least OpenAI if not midjourney. and that they accidentally sent a bunch of shit they shouldn't, like the contents of deleted blogs (trust and safety still has access to those. they are not fully purged)

i do not trust that these partners would delete data retroactively when asked. if they've already fed it in for training, there's no way to undo that.

you should be able to opt out in the future, but i wouldn't trust tumblr to be able to safely truncate reblog chains and not provide that data that way. i mean, andrew spittle clearly made a call that he did not have a duty of care to anyone who took the terms of service in good faith, so i would not trust him on anything in the future.

in terms of public posts, art or otherwise, the difference is between "your stuff is maybe being scraped if it's publicly available" and "your stuff is definitely being scraped if it's publicly available". the private stuff is a bigger deal bc that should have in no way come with the expectation that someone could have fed it to AI. private asks, blogs, and deleted content should never have been available.

tumblr in general has a pretty poor understanding of how LLMs and other generative AI work. in particular art/image AI generators do not store a copy of your image/trace it/paste it into things. that would be way, way too much data for anyone to install and use. it learns from it, just like a human artist learns by looking at other people's art how color and edges work.

your work being fed into generative AI does not in any way make it more likely (especially if you are not an extremely prolific person with an extremely hard to replicate style) that people will personally bypass you as an artist to use AI to replicate your stuff. incredibly few people are good, popular, and important enough that that is an actual risk.

AI generators are a threat to creative industries as a whole in the same way machine translation threatened translators. it is something that cannot be solved by handwringing about how special your art or your feelings are. collective action (or, you know, ideally, completely destroying the current system) is the only way to prevent capitalism from adopting every new technology it can specifically in ways that involve treating workers badly.

we have a misinformation problem. we have a problem where google is getting less and less useful every day, because their algorithms are tweaked to make them money and not to be a good search engine, and because people who are paid to get useless shit to the top of the results have way more time to work on that than people writing actual useful articles.

https://www.404media.co/tumblr-and-wordpress-to-sell-users-data-to-train-ai-tools/

Tumblr and Wordpress are preparing to sell user data to Midjourney and OpenAI, according to a source with internal knowledge about the deals and internal documentation referring to the deals. 

The exact types of data from each platform going to each company are not spelled out in documentation we’ve reviewed, but internal communications reviewed by 404 Media make clear that deals between Automattic, the platforms’ parent company, and OpenAI and Midjourney are imminent.

The internal documentation details a messy and controversial process within Tumblr itself. One internal post made by Cyle Gage, a product manager at Tumblr, states that a query made to prepare data for OpenAI and Midjourney compiled a huge number of user posts that it wasn’t supposed to. It is not clear from Gage’s post whether this data has already been sent to OpenAI and Midjourney, or whether Gage was detailing a process for scrubbing the data before it was to be sent. 

Gage wrote:

“the way the data was queried for the initial data dump to Midjourney/OpenAI means we compiled a list of all tumblr’s public post content between 2014 and 2023, but also unfortunately it included, and should not have included:

- private posts on public blogs - posts on deleted or suspended blogs - unanswered asks (normally these are not public until they’re answered) - private answers (these only show up to the receiver and are not public) - posts that are marked ‘explicit’ / NSFW / ‘mature’ by our more modern standards (this may not be a big deal, I don’t know) - content from premium partner blogs (special brand blogs like Apple’s former music blog, for example, who spent money with us on an ad campaign) that may have creative that doesn’t belong to us, and we don’t have the rights to share with this-parties; this one is kinda unknown to me, what deals are in place historically and what they should prevent us from doing.”

Gage’s post makes clear that engineers are working on compiling a list of post IDs that should not have been included, and that password-protected posts, DMs, and media flagged as CSAM and other community guidelines violations were not included.

Automattic plans to launch a new setting on Wednesday that will allow users to opt-out of data sharing with third parties, including AI companies, according to the source, who spoke on the condition of anonymity, and internal documents. A new FAQ section we reviewed is titled “What happens when you opt out?” states that “If you opt out from the start, we will block crawlers from accessing your content by adding your site on a disallowed list. If you change your mind later, we also plan to update any partners about people who newly opt-out and ask that their content be removed from past sources and future training.” 

404 Media has asked Automattic how it accidentally compiled data that it shouldn’t share, and whether any of that content was shared with OpenAI, but did not immediately hear back from the company. 404 Media asked Automattic about an imminent deal with Midjourney last week but did not hear back then, either.

Another internal document shows that, on February 23, an employee asked in a staff-only thread, “Do we have assurances that if a user opts out of their data being shared with third parties that our existing data partners will be notified of such a change and remove their data?”

Andrew Spittle, Automattic’s head of AI replied: “We will notify existing partners on a regular basis about anyone who's opted out since the last time we provided a list. I want this to be an ongoing process where we regularly advocate for past content to be excluded based on current preferences. We will ask that content be deleted and removed from any future training runs. I believe partners will honor this based on our conversations with them to this point. I don't think they gain much overall by retaining it.” Automattic did not respond to a question from 404 Media about whether it could guarantee that people who opt out will have their data deleted retroactively.

News about a deal between Tumblr and Midjourney has been rumored and speculated about on Tumblr for the last week. Someone claiming to be a former Tumblr employee announced in a Tumblr blog post that the platform was working on a deal with Midjourney, and the rumor made it onto Blind, an app for verified employees of companies to anonymously discuss their jobs. 404 Media has seen the Blind posts, in which what seems like an Automattic employee says, “I'm not sure why some of you are getting worked up or worried about this. It's totally legal, and sharing it publicly is perfectly fine since it's right there in the terms & conditions. So, go ahead and spread the word as much as you can with your friends and tech journalists, it's totally fine.”

Separately, 404 Media viewed a public, now-deleted post by Gage, the product manager, where he said that he was deleting all of his images off of Tumblr, and would be putting them on his personal website. A still-live post says, “i've deleted my photography from tumblr and will be moving it slowly but surely over to cylegage.com, which i'm building into a photography portfolio that i can control end-to-end.” At one point last week, his personal website had a specific note stating that he did not consent to AI scraping of his images. Gage’s original post has been deleted, and his website is now a blank page that just reads “Cyle.” Gage did not respond to a request for comment from 404 Media. 

Several online platforms have made similar deals with AI companies recently, including Reddit, which entered into an AI content licensing deal with Google and said in its SEC filing last week that it’s “in the early stages of monetizing [its] user base” by training AI on users’ posts. Last year, Shutterstock signed a six year deal with OpenAI to provide training data.

An ex-colleague of mine mentioned yesterday that there may be contacts between Automattic and midjourney in that direction, but nothing is public yet and I don't have any more info. They probably won't have anything specific to share either, since they left the company weeks ago too. That being said:

I have no reason to doubt my ex-coworker word, they are a trustworthy person.

Tumblr's CEO has been absurdly enthusiastic (comically, even) about AI, and is a big fan of LLMs and 'AI' companies.

A deal with midjourney could solve tumblr financial issues (not the same company, but openAi is paying up to 5 million/year to news companies to use their content as training data... tumblr generates several orders of magnitude more content than any newspaper or any media company and it only would need a 20 to 30 million per year deal to be profitable)

I won't be posting chapter excerpts on socials for these last sets of chapters. The spoilers are heavy. But, here's the wonderful cover that @binturong-draws made for the book. Haven't posted it here in a while.