To help you stay on top of the latest news, our AI practice group has compiled a roundup of the developments we are following.
- The New York Times reports that AI company Anthropic will settle a copyright infringement lawsuit brought by a group of authors and publishers against the company for $1.5 billion, the largest such sum in U.S. history for a copyright case. The authors and publishers argued that Anthropic’s use of their copyrighted works to train its Claude AI chatbot infringed on their copyright. In a ruling in June, as explained by MBHB Partner Michael Borella, Judge William Alsup found in a ruling for summary judgment that the general use of copyrighted content to train AI models fell under the doctrine of fair use, though only through the use of digital versions of purchased books. However, he also found that Anthropic’s use of pirated ebooks for training did not constitute fair use. As part of the settlement, Anthropic will pay $3,000 per work to over 500,000 authors, as well as state that they did not use pirated works for their publicly released AI technology, and to delete the pirated works the company has already downloaded. Judge Alsup still needs to approve the settlement for it to take legal effect, and if approved will represent the end of one of the first and most high profile legal cases in the U.S. addressing the AI-copyright issue, and will likely have further effects and inform other legal decisions on the topic in the future.
- Bloomberg reports on a new copyright infringement lawsuit filed this past week by media conglomerate Warner Bros. Discovery against AI company Midjourney. In the complaint filed in a federal district court in California, Warner Bros. alleges that Midjourney’s eponymous AI image generation product can create images and videos depicting copyrighted characters, including Superman, Batman, Wonder Woman, Bugs Bunny, and Scooby-Doo, without the authorization of the studio, which is asking for damages of up to $150,000 per act of infringement. This is not the first copyright infringement suit to be brought against Midjourney this year; as reported by this AI Roundup in June, both the Walt Disney Company and Comcast’s NBCUniversal sued Midjourney on similar grounds. Midjourney, along with other AI companies, have argued that the use of copyrighted content to train AI models falls under the “fair use” in U.S. copyright law, though few legal decisions on the matter have been made. The case is Warner Bros. Entertainment Inc. v. Midjourney Inc., 2:25-cv-08376, U.S. District Court, Central District of California (Los Angeles).
- The Financial Times reports that OpenAI will begin mass production of its own AI chips in 2026. The chips, dubbed as “XPUs” and designed in partnership with Broadcom, is aimed to help lessen dependence on AI chips from industry leader Nvidia. Several other companies, including Amazon, Google, and Meta, have each designed their own AI chips to help train and run their AI models, the process of which consumes vast amounts of computing power. OpenAI’s collaboration with Broadcom began last year, though the timeline for mass production was only confirmed this past week, as Broadcom CEO Hock Tan told investors that an unnamed company had committed to $10 billion dollars of custom AI chips, and sources told the FT that OpenAI was the mystery client, though both companies declined to comment.
- The South China Morning Post reports on the implementation of a new law in China that requires the labeling of all AI-generated content on the internet. Social media platforms, including WeChat, Douyin (the Chinese domestic equivalent of TikTok), microblog site Weibo, and RedNote all announced new features to comply with the law. The law, which was passed in March but rolled out this past week, “requires explicit and implicit labels for AI-generated text, images, audio, video and other virtual content. Explicit markings must be clearly visible to users, while implicit identifiers — such as digital watermarks — should be embedded in the metadata.” The Chinese government has recently increased its oversight of the AI industry, which has rapidly grown in China, and the new law was drafted in part by the Cyberspace Administration of China (CAC), which has laid out goals to strictly monitor AI content and enforce the labeling requirements.
- WIRED reports on the rise of AI agents being used for sports gambling. Since the U.S. Supreme Court struck down bans on sports gambling in 2018, the industry has rapidly grown, fueled by sports-betting smartphone apps, with over $150 billion spent on sports wagers in 2024. Several startups have emerged, claiming to use generative AI to give their customers an edge in their bets, with one, MonsterBet, charging $77 per month for its model that projects sports results based on information scraped from the Internet. Some AI systems are in development that can place wagers themselves on behalf of users, though as agents generally cannot access traditional bank accounts, these systems often make use of cryptocurrency. This technology is still at a very early stage, and there is little evidence that such products actually improve the odds of winning bets, but efforts to combine AI, gambling, and in some cases crypto are likely to continue for the foreseeable future.
[View source.]
