Background
Freelance illustrator Robert Santora was contracted by Hachette Book Group to design original cover artwork for ten novels by bestselling romance and thriller author Sandra Brown. Years later, Hachette undertook a reprint campaign and redesigned the covers. Santora alleged that rather than commissioning entirely new artwork, Hachette used artificial intelligence to alter and reproduce his copyrighted illustrations, creating derivative works without his authorization or compensation.
Hachette denied that any AI was involved, maintaining that the replacement covers were designed by its in-house (human) design team and were independently created, not derived from Santora’s prior work. The central dispute became whether the new covers were sufficiently similar to Santora’s originals to constitute infringement.
The Court’s Holding
The court granted Hachette’s motion and dismissed the claims. Applying the substantial similarity standard for copyright infringement, the court found that Santora had not demonstrated the requisite similarity between his original cover art and the accused Hachette covers. The dismissal appears to have been granted without prejudice (Law360 noted the “for now” qualifier), leaving Santora the option to amend his complaint or appeal.
The court’s ruling was grounded in the traditional substantial similarity analysis rather than on the AI question directly: even accepting Santora’s allegations about AI use as true for the purposes of the motion, he could not clear the threshold of showing his protected expression was reproduced in the new covers. The ruling does not reach whether AI-generated works can infringe or whether AI reproduction constitutes derivative-work creation under copyright law—those questions remain open for a case where the similarity is established.
Key Takeaways
- The threshold substantial similarity requirement is not relaxed simply because AI is alleged to have been the infringement mechanism. A plaintiff must still identify which specific protectable expression was copied and show it appears in the accused work.
- The “for now” framing of the dismissal suggests the court left room for a more developed factual record. If Santora or similarly situated artists can obtain and analyze the AI prompts, training data, or intermediate outputs used to create allegedly infringing covers, they may be able to establish the similarity that was missing here.
- Cases like this illustrate why technical AI forensics (prompt analysis, model attribution, style similarity metrics) may become standard tools for AI copyright plaintiffs going forward.
- Hachette’s defense—that human designers, not AI, created the covers—underscores that AI copyright disputes will frequently involve contested facts about the creative process that are difficult to resolve at the pleadings stage.
Why It Matters
As AI image generation tools become standard in commercial design workflows, cases like Santora v. Hachette are the leading edge of a wave of disputes between legacy contractors (human artists who created original works) and corporations that may be using AI to cheaply replace or reinterpret that work. The substantial similarity doctrine was developed for a world where human copying was the infringement mechanism; courts are now adapting it to the reality that AI can produce outputs with stylistic resemblance to training examples without performing traditional copying.
The case is a cautionary note for artists: claiming AI was used, without evidence of what the AI output actually reproduces from the original, is not enough to survive a motion to dismiss. The next wave of cases will likely turn on plaintiffs’ ability to use discovery—email records, AI system logs, prompt histories—to show what specific elements were fed into the AI and what came out.