AI News Roundup – California legislature passes new AI regulation bill, proposed Senate bill would allow waivers from federal AI regulation during testing phase, Albania welcomes an AI chatbot into its cabinet, 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.

• • POLITICO reports on the passage of a new AI regulation bill by the California legislature. The bill, a new incarnation of a bill that was vetoed by Governor Gavin Newsom last year (as this AI Roundup covered), would require AI companies to disclose AI safety testing procedures and certify with the state that they are being adhered to, with lesser financial penalties than last year’s bill. Newsom vetoed the previous bill as restricting innovation in the AI space, but his office has yet to indicate what he will do for the newest bill. California’s legislature has also considered many other AI regulation bills this session, and has sent several to Newsom’s desk as well, including a bill to require safeguards for AI chatbots that interact with children, following controversy around the harms of “AI companions.” Newsom has until October 12 to sign or veto the bills.
• • Bloomberg reports on a draft bill in the U.S. Senate that would loosen federal regulations on AI during development and testing. The bill, dubbed the “Strengthening Artificial intelligence Normalization and Diffusion By Oversight and eXperimentation Act,” or SANDBOX Act, was introduced by Texas Republican Senator Ted Cruz earlier this week. The bill directs the White House Office of Science and Technology Policy to establish a “regulatory sandbox” program that would waive federal AI regulations during the development and testing of AI models for a two-year period, renewable up to eight years. AI companies in the U.S. have generally opposed federal regulation of the technology, often citing the need to compete with foreign AI rivals such as China. Indeed, earlier this year, Cruz introduced an amendment in the Trump administration’s “One Big Beautiful Bill” that would have placed a moratorium on state-level AI regulation, though that proposal was eventually voted down. Cruz’s present bill was referred to the Senate’s Committee on Commerce, Science, and Transportation and is awaiting further legislative action.
• • BBC News reports on Albania’s introduction of the world’s first “AI minister,” giving an AI chatbot a place in the nation’s cabinet. Albania Prime Minister Edi Rama introduced “Diella,” an AI chatbot, as a new member of his cabinet as minister for public procurement. The move is intended to crack down on corruption in the public procurement process, with Rama stating that, with Diella’s help, “not only will we wipe out every potential influence on public biddings — we will also make the process much faster, much more efficient and totally accountable.” The Albania government had already been using Diella (as an AI-powered virtual assistant for aiding applicants with government documents), but the new role is unprecedented. Albania’s Democratic Party, currently in opposition to Rama, called the move “unconstitutional” and “ridiculous,” though some specialists are optimistic about the potential for AI to be used to investigate public corruption.
• • The CBC reports on a major report on education policy prepared by the Canadian province of Newfoundland and Labrador that included over a dozen citations to nonexistent sources, signaling the uncareful use of AI to prepare it. The report, which calls for the province’s government to “provide learners and educators with essential AI knowledge, including ethics, data privacy, and responsible technology use,” references a nonexistent 2008 movie that was provided in a university style guide, from which it was likely introduced into an AI model. The two co-chairs of the group that prepared the report declined to comment.
• • The Financial Times reports on a major investment initiative by OpenAI and Nvidia into the U.K. to be announced during Donald Trump’s state visit to the country this coming week. The two companies, whose CEOs will join Trump on the visit, are expected to announce support for the construction of AI data centers in the country. While the full details are still unknown, people with knowledge of the plans told The Financial Times that OpenAI and Nvidia will provide software and hardware respectively for the projects, while the U.K. government will provide the energy, though they cautioned that the plans are not finalized. The companies, the White House, and the U.K. government each declined to comment or did not respond.