When users first learned of Adobe’s new terms of service (which were quietly updated in February), there was quite a stir. Adobe told users that it could access their content “through both automated and manual methods” and use “techniques such as machine learning to improve (Adobe’s) software and services.” Many understood the update to mean that the company was forcing users to grant unlimited access to their work, in order to train Adobe’s generative AI: Firefly.
Late Tuesday, Adobe issued a clarification: In a updated version of your terms of service agreementpledged not to train AI on its users’ content stored locally or in the cloud and gave users the option to opt out of content analysis.
trapped in the crossfire of intellectual property lawsuits, the ambiguous language used to update the terms above sheds light on a climate of acute skepticism among artists, many of whom rely too much on Adobe for their work. “They already broke our trust,” says Jon Lam, senior storyboard artist at Riot Games, referring to how award-winning artist Brian Kesinger discovered images were generated in the style of his art and sold under his name on his stock image site, without his consent. Earlier this month, the late photographer’s estate Ansel Adams publicly scolded Adobe for allegedly selling generative AI knockoffs of his work.
Scott Belsky, Adobe’s chief strategy officer, attempted to calm concerns when artists began protesting. clarifying that Machine learning refers to the company’s non-generative artificial intelligence tools: Photoshop’s “Content Aware Fill” tool, which allows users to seamlessly remove objects in an image, is one of many tools done by machine learning. But while Adobe insists that the updated terms do not give the company ownership of the content and that they will never use user content to train Firefly, the misunderstanding triggered a broader discussion about the company’s market monopoly and how a change as this could threaten the livelihoods of artists at any point. Lam is one of the artists who still believes that, despite Adobe’s clarification, the company will use work created on its platform to train Firefly without the creator’s consent.
Nervousness over non-consensual use and monetization of copyrighted works using generative AI models is not new. Early last year, artist Karla Ortiz was able to generate images of her work using her name in several generative AI models; a crime that gave rise to a class action lawsuit against Midjourney, DeviantArt and Stability AI. Ortiz was not alone: Polish fantasy artist Greg Rutkowski discovered that his name was one of the most used indications in Stable Broadcast when the tool was first released in 2022.
As the owner of Photoshop and creator of PDF files, Adobe has reigned as the industry standard for over 30 years, powering the majority of the creative class. An attempt to acquire a product design company. Figma was blocked and abandoned in 2023 by antitrust concerns that attest to its size.
Adobe specifies that Firefly is “ethically trained” on Adobe Stock, but longtime stock image contributor Eric Urquhart insists that “there was nothing ethical about how Adobe trained the AI for Firefly,” noting that Adobe does not own rights. to any image of individual contributors. Urquhart originally posted the images of himself on Fotolia, a stock image site, where he agreed to licensing terms that did not specify any use for generative AI. Fotolia was then acquired by Adobe in 2015, which implemented silent terms of service updates that then allowed the company to train Firefly using Photos of Eric without his explicit consent: “the language in the current TOS change is very similar to what I saw in the Adobe Stock TOS.”