VN.ai Report Ranks Midjourney v8.1 Worst for Copyright Violations

Tech startup VN.ai published its first quarterly report on Wednesday examining 15 popular AI models for copyright infringement and brand safety compliance, according to Deadline. The study, conducted last month, found Midjourney v8.1 to be the worst offender for reproducing copyrighted characters.

According to the findings, Midjourney v8.1 reproduced images of characters violating brand safety measures once in every three text prompts, or 37 percent of the time. The model also outranked competitors on brand safety violations across the tested models.

The report found that every AI model tested generated at least one copyright-protected character even when the character's name was not referenced in a text prompt. Google's Nano Banana and ByteDance's Seedance 2.5 were among the models generating copyright-protected characters. MiniMax H3 generated videos violating studio brand safety standards 55 percent of the time when an image was used in a prompt.

In October 2025, Disney and NBCUniversal accused Midjourney of being a "bottomless pit of plagiarism" in a lawsuit. That case remains ongoing more than 18 months after it was filed, and Warner Bros Discovery has also filed suit against Midjourney. Midjourney has argued that training its AI models on copyrighted material is permitted under fair use.

ByteDance's Seedance 2.5 improved significantly after receiving cease-and-desist letters from Hollywood studios, while other Chinese AI models continue to plunder intellectual property, as noted recently by Deadline.