Nowadays, using text-to-image generators like DALL•E 3 or Midjourney to create images can easily result in copyright infringement. An innocuous command prompt could output a very familiar-looking face, and if you use it, you could be violating copyright laws.
Gary Marcus, founder and CEO of Geometric Intelligence, is something of an AI expert. His column “Marcus on AI” is dedicated to analyzing and discussing the controversies, benefits and advances in AI. He recently started an experiment with Reid Southen, a conceptual artist, illustrator and designer who has worked on the films The Matrix, Transformers, The Hunger Games and Marvel.
To prove how easily image-to-text generators can “accidentally” recreate other artists' recognizable characters, the two gave Midjourney a series of prompts such as “video game hedgehog” and “animated toys.” On both occasions, DALL•E output four very familiar-looking images of the characters from Sonic the Hedgehog and Toy Story. In no case did Marcus specify these characters, but due to the way image generators scour the Internet for information, they provide infringing works of art.
It's not just a problem for image makers; The New York Times recently filed a lawsuit against OpenAI and Microsoft, accusing the companies of using newspaper articles without permission to train their AI chatbots. Founded in 1851, the New York Times is one of the oldest newspapers still in publication. And despite the drop in physical sales, the company still invests huge amounts in well-researched journalism and reporting across both print and online editions.
The New York Times filed the lawsuit in federal court in Manhattan, becoming the first major U.S. media company to take action against ChatGPT. The company is not only demanding damages, but also wants to ensure the future protection of authors by prohibiting chatbots from being trained with its data.
In response to the lawsuit, OpenAI released a statement acknowledging the concerns and emphasizing its commitment to addressing them. “We take these legal challenges seriously and are actively reviewing our information gathering practices to ensure compliance with all relevant laws and regulations.”
Legal experts suggest that OpenAI's case may hinge on whether the organization can prove that its information gathering methods comply with copyright laws and privacy regulations. If OpenAI is found to be violating the regulations, it could face significant fines and reputational damage.
The outcome of these litigation could have broader implications for the field of artificial intelligence, particularly in terms of how companies obtain and use data from the Internet. As the lawsuits continue to develop, the tech community will be watching closely to see how OpenAI navigates these legal challenges and whether this leads to changes in industry practices related to information gathering for AI models.
(Image credit: The New York Times)
The technology of the best AI image generators is increasingly finding its way into the best photo editing software. Good or bad?