3 Count: Spider Leak

Have any suggestions for the 3 Count? Let me know via Bluesky, Mastodon or LinkedIn.

1: Sony Battling Spider-Man: Brand New Day Post-Credit Scene Leaks, Scrubbing Social Media Via Copyright Claims

First off today, Tom Phillips at IGN reports that Sony has been scrambling to file takedown notices against a major leak pertaining to the upcoming film, Spider-Man: Brand New Day.

The film, slated to be released on July 31, features a post-credits scene that has leaked online prior to the film’s release. Sony has been aggressively filing takedown notices against social media accounts that have shared the scene, however, many copies of it remain online.

It is unclear where the leak came from though the video being shared appears to have been captured by someone in a theater. Most likely, theaters were given early access to the film and one of their employees recorded and leaked the scene.

2: A German Court Says an AI Comic Made From a Dog Photo Doesn’t Infringe Copyright

Next up today, Alex Cooke at Fatoppers reports that a German appeals court has ruled that an AI comic made from a copyright-protected photo does not automatically infringe on the copyright of the photographer.

The lawsuit was filed by a photographer who specializes in underwater dog portraits. He claimed that the defendant used his photo to create an AI-generated image based on his work. The lower court ruled against the photographer, saying that an AI image was a permitted “free adaptation” under German copyright law.

The appeals court disagreed with that finding but still found no direct infringement. That’s because, according to the court, the AI transformed the original photo, altering elements such as framing, lighting, and angle. Simply being able to recognize the source material does not, by itself, mean that the AI image is an infringement.

3: AI Companies Are Buying Antique Books, Ingesting Their Contents to Train Models, and Then Destroying Them at Incredible Scale

Finally today, Frank Landymore at Futurism reports that AI companies have been buying large numbers of books, including antique books, to train their models. As part of that process, they are destroying the books, removing their spines for easier scanning and then destroying the rest.

The move is copyright-related. A recent decision found Anthropic liable for copyright infringement not due to AI training, but for the use of pirated books. However, there is no copyright violation in purchasing legally-obtained books and then destroying them. This is covered under the doctrine of first sale.

The move is being aided by ISBNdb, which is a database of books and their ISBNs. They help AI companies purchase anywhere from 1,000 to 1 million books, with the intent that the books will be scanned and then used to train AI systems.

The 3 Count Logo was created by Justin Goff and is licensed under a Creative Commons Attribution License.