Adobe, the corporation that makes widely used creative software like Photoshop, unveiled earlier this year an update to the fine print of its Terms of Service that angered many creators who rely on their platform. The added language was vague and expansive but appeared to indicate that Adobe had granted itself permission to access any content—published or not—that users had created to train its new machine learning AI.

TOS changes are one-sided, giving corporations significant power to impose changes on their use policies. Language in the new TOS would allegedly force the digital artists who rely on Adobe’s products to hand control over their work to a corporation (see Standard Form Contracts). Generative AI’s, which produce “novel” content, train themselves on huge datasets, usually comprised of publicly available works. That process has led to concerns, such as those voiced here, that generative AI could be reproducing the intellectual property of artists and creators without obtaining their permission.

Critics of the TOS change alleged that Adobe tried to lure users into consenting to broad use of their work through the placement of a clause granting itself permission in a TOS agreement. Users were outraged at the changes in the fine print, so Adobe first tried to clarify the clause (which they failed to do) and then committed to changing the language, promising to add language that prohibits Adobe from using the users’ works to train generative AI. The whole saga demonstrates the unilateral, imbalanced power corporations can wield—and the importance of consumers reading and speaking out about unfair terms in the fine print of their contracts.