Let's start here. You are angry at the wrong thing.
You watched generative AI arrive and decided the tool was the villain. "AI steals." "AI is theft." You posted it on Instagram. You tweeted it on X. You said it on Reddit. You said it on the very platforms that had already taken your work and sold it to the companies you're mad at. And you did it because you didn't read what you agreed to when you signed up.
That's not an opinion. That's a receipt.
The Fine Print You Didn't Read
Security.org took it further. They built a fake consent form that included a clause granting the company naming rights to the user's firstborn child. Only 16 people out of more than 1,000 caught it.1
This is who we're dealing with. People who can't be bothered to read the agreement they clicked "Accept" on before uploading their portfolio, their music, their writing, their photography, and their entire creative identity to platforms whose business model is monetizing that content.
You're an adult. You clicked "I Agree." Nobody held a gun to your head. You just didn't want to read.
What You Actually Agreed To
Here's what those agreements say. Not the AI-era updates. The language that existed years before generative AI was available to the public.
"When you share, post, or upload content that is covered by intellectual property rights on or in connection with our Products, you grant us a non-exclusive, transferable, sub-licensable, royalty-free, and worldwide license to host, use, distribute, modify, run, copy, publicly perform or display, translate, and create derivative works of your content."2
Read that again. Transferable. Sub-licensable. Worldwide. Create derivative works. That language doesn't say "to show your friends your vacation photos." It says Meta can do virtually anything with your content, hand it to anyone they want, and they don't owe you a dime.
"You grant Tumblr a non-exclusive, worldwide, royalty-free, sublicensable, transferable right and license to use, host, store, cache, reproduce, publish, display... create derivative works of, such Subscriber Content."
"You also agree that this license includes the right for Tumblr to make all publicly-posted Content available to third parties selected by Tumblr, so that those third parties can syndicate and/or analyze such Content on other media and services."3
That was 2014. Ten years before any of you were worried about AI. Tumblr told you they could give your content to third parties for analysis on "other media and services." You agreed. You kept posting. And in 2024, Tumblr's parent company Automattic struck deals with OpenAI and Midjourney to sell user content dating back to 2014 as AI training data.4
You handed them the keys and they drove the car exactly where the contract said they could.
"By submitting, posting or displaying Content on or through the Services, you grant us a worldwide, non-exclusive, royalty-free license to make your Content available to the rest of the world... including, for example, for use with and training of our machine learning and artificial intelligence models, whether generative or another type."5
X made it explicit in late 2024. No ambiguity. No loophole. Your posts train Grok. By January 2026, they expanded the definition of "Content" to include everything you input or generate, with no opt-out.6 The platform you used to complain about AI training is itself training AI on your complaints about AI training.
Reddit struck a deal with Google reportedly worth $60 million per year to license its user-generated content for AI training.7
Reddit CEO Steve Huffman said it plainly: "Reddit's corpus of data is really valuable. We don't need to give all of that value to some of the largest companies in the world for free."8 He wasn't talking about protecting users. He was talking about monetizing them.
And it wasn't just social media. When I worked for Zoho, their employment contract stated that they have rights to your likeness for perpetuity and can use them in marketing materials. This was before generative AI was available to the public.
As University of Houston law professor Jacqueline Lipton noted, platforms "likely weren't contemplating AI training when drafting their user agreements," but their language is "so broad that it's inconceivable that that's not in the set of rights they get."9
You gave your work to corporations with a contract that said they could do anything with it. They did something with it. And now you're mad at the tool they used instead of the contract you signed.
The Sale You Weren't Told About
This is the part that should actually make you angry.
The platforms didn't just benefit from broad licensing language. They actively packaged and sold your content to AI companies. Tumblr sold posts dating back a decade to OpenAI and Midjourney.4 Reddit sold your discussions to Google for $60 million a year.7 Meta trained its own AI models on public Instagram and Facebook posts, with an opt-out process so buried and ineffective that Meta admitted it "may still choose to process a user's information and use it to train an AI anyway."10
Not because AI stole it. Because you gave it to them under terms that allowed it. The anger is valid. The aim is wrong.
"AI steals" is a slogan. The platforms sold. The companies bought. The users signed. That's not theft. That's commerce you weren't paying attention to.
The Cost of Ignorance
Here is the part nobody wants to hear.
If you are an artist who posted your work on Instagram, Tumblr, X, Facebook, Reddit, DeviantArt, or any other platform without reading the terms of service, you are not a victim of AI. You are a victim of your own lack of diligence. You signed a contract. You didn't read it. The contract said the platform could sublicense your work worldwide, create derivative works from it, and transfer those rights to third parties. The platform exercised those rights. You got mad at the wrong entity.
This doesn't mean the system is fair. It isn't. Terms of service are intentionally long, dense, and written in language designed to discourage reading. As a FindLaw survey found, "courts have tended to uphold such agreements whether the person actually read them or not."11 The deck is stacked. But that's an argument for better regulation and personal education. It is not an argument for screaming "AI steals" on the very platforms that sold your work.
Education matters. If you're going to have an opinion about AI, have an informed one. Know what you signed. Know who profited. Know the difference between a tool and the companies wielding it. Because right now, the loudest voices in this conversation are the least educated about what actually happened. And y'all sound like spoiled, entitled children throwing a fit.
The loudest voices in this conversation are the least educated about what actually happened. Educate yourself. Read what you sign. Aim your anger at the companies that sold your work, not the tool they sold it to.
Where The Energy Should Go
If you're an artist and you're angry, good. You should be. But channel that energy somewhere productive.
Push for regulation that requires platforms to get explicit, informed consent before licensing user content to third parties for AI training. Push for compensation structures that give creators a share when their work is used in training datasets.
And while you're at it, learn the tool. Because while you've been protesting AI, other artists have been using it. They've been learning how it works, what it can do, and how to make it serve their creative vision rather than replace it. The ones who survive this shift won't be the ones who screamed the loudest. They'll be the ones who adapted the fastest.
Darkframe was built using the same tool that displaced us. We didn't ask for permission. We didn't wait for the industry to come around. We directed the tool with decades of creative experience and built something that didn't exist before. We catch the same hate from artists who think using AI makes us traitors. But they can't tell the difference between the tool and the companies that trained it. They don't know what they signed. They don't know who sold their data. They just know they're angry and AI is the easiest target.
That's not a position. That's a tantrum.
Read your contracts. Educate yourself. Aim your anger at the right target. And if you've got creative energy to burn, for the love of god, make something with it instead of wasting it on a fight you already lost in the fine print.
Sources
- Security.org, "Americans Get an F on Digital Privacy Knowledge." Study of 1,000+ participants; bogus consent form included naming rights to firstborn child.
- Meta Terms of Service / Instagram Terms of Use. Content licensing provision. Verified via Facebook.com/terms and Facebook Help Center.
- Tumblr Terms of Service, 2014 (archived via Wayback Machine at web.archive.org). Third-party content sharing clause.
- 404 Media, "Tumblr and WordPress to Sell Users' Data to Train AI Tools," February 27, 2024. Deals with OpenAI and Midjourney for content dating back to 2014.
- X (formerly Twitter) Terms of Service, effective November 15, 2024. AI/ML training language added October 2024. Reported by CNN, CNBC, The Register.
- X Terms of Service update, effective January 15, 2026. Expanded "Content" to include user inputs and outputs. Reported by tesorb.com and Platform Policy Watch.
- Google-Reddit content licensing deal, $60 million annually. Reported by Bloomberg, Reuters, Engadget, AP, February 2024.
- Steve Huffman, Reddit CEO. Quoted in multiple outlets regarding Reddit's API monetization and AI licensing strategy.
- Bloomberg Law, "Google-Reddit AI Deal Heralds New Era in Social Media Licensing." Quotes University of Houston copyright law professor on breadth of platform licensing language.
- Helen Nissenbaum, Cornell Tech, quoted in Harvard Business Review. Referenced in PopSci, "Angry Instagram Posts Won't Stop Meta AI From Using Your Content."
- FindLaw.com, "Most Online Shoppers Don't Read the Fine Print," February 2015. Courts uphold click-wrap agreements regardless of whether user read them.