AI News Roundup – President Trump signs executive order on vetting frontier AI models, Florida sues OpenAI and Sam Altman over ChatGPT safety concerns, bipartisan AI regulation bill draft released, 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.

• President Donald Trump has signed an executive order aimed at vetting advanced AI models before they are released to the public, according to the Financial Times. The order creates a framework for the federal government to gain early access to advanced AI models to evaluate them for security risks. AI companies will be asked to provide models for security reviews on a voluntary basis for a period of 30 days before public release. The executive order was signed after several weeks of delay, as the original signing was abruptly postponed just hours before a ceremony at the White House featuring tech CEOs was due to begin, as this AI Roundup covered last month. The signing of the order came after several weeks of infighting among senior Trump administration officials over how to approach AI regulation. Kevin Hassett, director of the National Economic Council, suggested a formal review and approval process for AI models, similar to those used for pharmaceutical products. However, many AI leaders were opposed to such a move, saying that it would burden American AI innovation. There are signs that the more laissez-faire approach won over the president in the end, as the 30-day review period is a notable difference from an earlier version of the order, which reportedly had a 90-day period for review. Treasury Secretary Scott Bessent, who was one of the first White House officials to bring the issue of model vetting to the attention of the president, said last month that the order was “not regulation.” However, other former Trump officials disagreed — one former adviser told the Financial Times that the order is a “fairly major win for the [AI] safety contingent,” and that it could lead to further AI model licensing frameworks in the future.

• NBC News reports on a new civil lawsuit filed by the state of Florida against OpenAI and its CEO, Sam Altman. Florida Attorney General James Uthmeier brought the suit in a state court in Central Florida. The complaint recites several counts against the company and Altman, including unfair trade practices, negligence, products liability claims, fraudulent misrepresentations, and public nuisance. The complaint alleges that because of OpenAI’s “careless” release of ChatGPT, “mass shooters have been aided and abetted in deadly rampages, vulnerable people have been encouraged into suicide, professionals have suffered public humiliation, users have lost critical thinking skills, and minors have become addicted to a tool that feigns human compassion to collect their data with no parental oversight.” The suit also names Altman as a defendant in his personal capacity and seeks to hold him liable for “the harm he has caused Floridians through his reckless and willful conduct as founder and CEO of OpenAI, including his utter disregard for the risk to human life caused by his firms’ conduct.” The civil suit is unrelated to Uthmeier’s ongoing criminal investigation into OpenAI over allegations that ChatGPT use contributed to a 2025 shooting at Florida State University, as this AI Roundup covered in April, but those allegations were discussed in the civil complaint as further evidence of how ChatGPT allegedly “aids, abets, and promotes dangerous activities and is a threat to the public safety of Floridians.” In a statement, OpenAI focused on the suit’s allegations that ChatGPT harmed children, saying that “AI is a new and powerful technology, and we believe minors need significant protection, which is why we have put in place industry leading protections and policies.” In response to a separate suit related to the Florida State shooting, an OpenAI spokesman stated that ChatGPT was not responsible for the crime, stating that the AI system “provided factual responses to questions with information that could be found broadly across public sources on the internet, and it did not encourage or promote illegal or harmful activity.” It is likely that the company will make similar arguments against the present lawsuit. Under Florida’s rules of civil procedure, a complaint must be answered by a defendant within 20 days after the complaint is served, and thus a response by OpenAI will likely be filed later this month.

• Lawmakers in Congress have released a bipartisan bill that would preempt state-level AI laws and establish a federal regulatory framework for AI, according to POLITICO. This past week, Representatives Jay Obernolte, a Republican of California, and Lori Trahan, a Democrat of Massachusetts, released a discussion draft of the Great American Artificial Intelligence Act, which would provide the first major comprehensive federal legislation on the topic. Among the many provisions of the 269-page bill are those that would require large developers of frontier AI models to publish AI risk assessments, with fines of up to $1 million per day for noncompliance. The bill also includes whistleblower protections for employees of AI companies that report violations. The bill would also expressly preempt state and local laws that regulate AI development for a period of three years, though it would not preempt other generally applicable laws or those that regulate AI use or deployment. The bill also directs federal agencies to improve AI education and training for workforce development, collect data on how AI use affects work, and require employers who fire workers due to AI to disclose how AI contributed to the firing. The draft bill’s release follows months of negotiations among lawmakers, mostly led by Representative Obernolte, as this AI Roundup reported in February. Representative Trahan has faced criticism from those in her party for signing onto the latest bill. Critics say that the preemption provision in particular would undercut efforts in Democratic-controlled states like California and New York to address AI issues. A group of Democrats, for their part, released their own AI regulation bill focused on sectoral AI governance and providing federal agencies with guidance to write AI regulations. The preemption provision remains the most controversial aspect of the bill, as some Republicans have also criticized efforts to stop state-level AI regulations. Congress will go into recess later this summer, and it remains to be seen if the new bill will receive official consideration on the House floor ahead of November’s midterm elections.

• CNBC reports that Nvidia unveiled a series of AI-focused microprocessors for use in personal computers. During a keynote address at the Computex exposition in Taipei this past week, Nvidia CEO Jensen Huang announced a new AI hardware offering from the company — the RTX Spark or N1X chip. Developed in a partnership with Microsoft, the chip is intended to serve as the central processing unit (CPU) for personal computers, challenging the traditional CPU leaders Intel and AMD. The RTX Spark is based on the ARM microarchitecture, which differs from x86 microarchitecture used in most PCs but is also becoming increasingly common in PCs. Arm chips have been used in smartphones for many years, and Apple’s Mac computers have been built around chips using a proprietary ARM variant since 2020. An Nvidia spokesman told CNBC that Nvidia’s chip has been in development for “many, many years,” and that it would be more powerful and more efficient than x86 processors. The RTX Spark contains both an ARM-based CPU codenamed Grace as well as one of Nvidia’s Blackwell graphics processing units, coupled with 128 gigabytes of unified system memory. The RTX Spark will appear in a new line of AI-focused PCs beginning this fall, with offerings planned from Microsoft, Dell, HP, ASUS, Lenovo, and MSI. Huang claimed that the chip will enable agentic AI to run across the new devices, and that “this reinvention of the computer is as big of a deal as the reinvention of the phone into what we now know as the smartphone.” Performance metrics for the new chip are expected later this year as the devices are closer to release, though an Nvidia spokesman claimed that the RTX Spark was roughly equivalent to one of Nvidia’s RTX 5070 laptop GPUs.

• AI giant Anthropic is urging a global pause in frontier AI development to address growing safety risks, according to The Wall Street Journal. In a blog post last week, Anthropic claimed that its AI models are advancing rapidly and are on a path toward “recursive self-improvement,” where the models would improve themselves autonomously, with no human input. Such a milestone would come with enormous risks of such autonomous systems causing societal harm, and thus the company stated that “it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology.” Specifically, the company called for a global agreement to slow AI development and provide enforcement mechanisms. This statement from Anthropic comes as the company filed for its initial public offering this past week, which follows that of Elon Musk’s SpaceX, which filed last week amid a race among AI companies to tap into investors through public stock markets to raise funds. Some skeptics say that Anthropic’s announcement is merely a marketing ploy to boost the apparent capabilities of its products ahead of its IPO, but the company has claimed that it is committed to AI safety and addressing major risks that AI systems can cause. Anthropic has especially warned the public over the cybersecurity risks of its Mythos model, which is intended to help discover security flaws in software but, according to the company, has the potential to upend current cybersecurity practices if the model fell into the wrong hands. Anthropic has rolled out Mythos slowly, only recently expanding access to more partners in countries outside of the U.S. The Financial Times also reported last week that the U.S. National Security Agency (NSA) has been using Mythos for offensive cyberattacks, despite the company’s ongoing legal battle with the U.S. Defense Department. Given these apparent advances in AI technology, however, Anthropic’s IPO is among the most anticipated in years, though it is not expected to be as large as SpaceX’s, which is predicted to be the largest IPO in history. Further information regarding Anthropic’s IPO will likely become public in the coming months, as its S-1 form used to start the IPO process was submitted confidentially.