Legal Pressure on AI
Legal challenges to generative AI are accelerating — U.S. copyright cases over model training have now passed 100, signaling growing uncertainty about whether training on copyrighted material is lawful. Creators are pushing back: YouTubers have sued Amazon alleging their videos were used without permission, and similar suits are proliferating across media types. At the same time OpenAI is fighting multiple fronts — shutting down its Sora product while its trademark fight with OverDrive continues — and the company has asked state attorneys‑general to probe Elon Musk even as Musk amended his lawsuit to seek the removal of Sam Altman and Greg Brockman and offer any damages to OpenAI’s nonprofit. (noah-news.com) (aol.com) (crainscleveland.com) (aol.com) (arstechnica.com)
More than 100 United States lawsuits now challenge the way generative artificial intelligence systems were trained, and most of them turn on one plain question: can a company copy books, songs, photos, or videos into a model without asking first. (noah-news.com) That fight used to sound abstract, but this week it landed on YouTube. H3H3 Productions, Mike Bury, and Brett Dolan sued Amazon in Seattle federal court, saying Amazon scraped their videos to train its Nova Reel video generator. (yahoo.com) The complaint says Amazon did not just watch public videos like a person with a browser. It says Amazon used bots, rotating internet addresses, and virtual machines to get around YouTube’s technical barriers and pull down large volumes of video. (thenextweb.com) That detail matters because copyright law and the Digital Millennium Copyright Act are two different gates. One asks whether Amazon copied protected work, and the other asks whether Amazon broke the digital locks while doing it. (yahoo.com) Courts are now being asked to decide whether training a model is closer to reading a library or photocopying the whole building. The United States Copyright Office said in its May 2025 report on generative artificial intelligence training that the law here is unsettled and highly fact-specific. (copyright.gov) OpenAI is fighting a different kind of case at the same time. OverDrive, the Cleveland company behind the student reading app Sora, sued in Ohio federal court in November 2025, saying OpenAI’s video product used the same name and created marketplace confusion. (crainscleveland.com) By April 8, 2026, Crain’s Cleveland reported that OpenAI had shut down the consumer-facing Sora website while the trademark case continued, even though the underlying video technology had not disappeared. (crainscleveland.com) Then there is the Elon Musk case, which is about control as much as money. On April 6, 2026, OpenAI asked the attorneys general of California and Delaware to investigate Musk and his associates for what it called improper and anti-competitive behavior tied to his effort to block OpenAI’s restructuring. (cnbc.com) A day later, Musk amended his own lawsuit. Ars Technica reported that he asked the court to remove Sam Altman and Greg Brockman from leadership roles and said any damages recovered should go to OpenAI’s nonprofit arm, not to him personally. (arstechnica.com) Put together, these cases hit three different pressure points at once. Copyright cases ask what data went in, the Sora trademark case asks what name can be sold, and the Musk fight asks who gets to steer one of the most powerful artificial intelligence companies in the world. (noah-news.com)
