AI News Roundup – U.N. tackles AI, Nvidia invests billions into OpenAI, DeepMind reveals newest robotics AI models, and more

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 on a new initiative for AI governance announced by the United Nations. During this past week’s U.N. General Assembly, the organization announced “a global dialogue on artificial intelligence governance,” aimed at collecting ideas and best practices from around the world on the topic. During a session of the General Assembly, several speakers exhorted the body to take action on AI, especially to ensure that the technology is not kept only in the hands of a few powerful countries or technology companies. However, the U.S. government has remained skeptical of such efforts, a White House technology official said this past week that “We totally reject all efforts by international bodies to assert centralized control and global governance of A.I.” No concrete U.N. actions have been yet taken, though an independent AI regulatory body has been proposed. U.N. Secretary General António Guterres said that the AI actions represented “a significant step forward in global efforts to harness the benefits of artificial intelligence while addressing its risks.” 
• • Reuters reports on a new $100 billion investment into OpenAI by AI chipmaker titan Nvidia. The move tightens links between perhaps the most influential hardware and software players in the AI industry. Nvidia will gain a stake in OpenAI, while OpenAI will gain cash as well as access to Nvidia’s most advanced AI chips. OpenAI CEO Sam Altman said in a statement that “compute infrastructure will be the basis for the economy of the future, and we will utilize what we’re building with Nvidia to both create new AI breakthroughs and empower people and businesses with them at scale.” OpenAI had previously been reported as developing its own AI chips, though a source told Reuters that the Nvidia deal did not affect those plans. 
• • The Financial Times reports on new AI models from Google DeepMind focused on robotics. This AI Roundup previously reported on two DeepMind AI models, Gemini Robotics and Gemini Robotics-ER, focused on using reasoning to control robotic arms. This past week, DeepMind revealed updated versions of both models—Gemini Robotics 1.5 and Gemini Robotics-ER 1.5. According to the company, a robot using the new models was able to plan out complex tasks, such as folding laundry and sorting it based on color. Another advancement in the new models, dubbed “motion transfer,” allows models to use skills designed for one type of robot, such as robotic arms, on other types, such as humanoid robots. DeepMind emphasizes that robotics-focused models still needed more development time before they could be introduced safely into human environments. 
• • The Verge reports on new developments in a copyright infringement lawsuit brought against AI music startup Suno by a group of record labels. In an amended complaint filed recently, the record labels allege that Suno pirated songs from YouTube to train its generative AI models. The labels say that this involved circumventing YouTube’s encryption programs and “has facilitated Suno’s ongoing and mass-scale infringement.” Suno’s training datasets have not been made public, though Suno, like other AI companies, has claimed that its use of copyrighted content for AI training falls under fair use. However, pirating content has been looked at skeptically by judges as part of this analysis, such as in the recent Anthropic AI settlement. Further developments in this lawsuit are expected over the coming months. 
• • CalMatters reports on record financial penalties placed on an attorney who submitted AI–hallucinated content in court filings. A Los Angeles attorney, who admitted to not reviewing AI-generated text that he included in an appeal filing, was fined $10,000. The court said that 21 of the 23 cases cited in the attorney’s brief were nonexistent, and restated that “no brief, pleading, motion, or any other paper filed in any court should contain any citations—whether provided by generative AI or any other source—that the attorney responsible for submitting the pleading has not personally read and verified.” U.S. courts have been cracking down on AI usage in court filings in recent months—California’s Judicial Council issued guidelines for judges to either ban generative AI entirely in their courts or adopt an AI use policy by the middle of December. The attorney involved in the case told CalMatters that he thought it was unrealistic to expect lawyers to stop using AI, but that he hopes his example “will help others not fall into the hole. I’m paying the price.”