Элон Маск проиграл суд по делу о прозрачности данных для обучения ИИ в Калифорнии

'Strains Credulity': Elon Musk Loses Crucial Round in Court Battle Over California's New AI Transparency Law As Elon Musk fights to keep xAI's training data under wraps, the real contest is over who gets to see how the algorithms shaping modern life actually work Elon Musk has suffered an early setback in his legal fight against California's new AI transparency rules, after a federal judge refused to halt enforcement of a state law that forces companies like his start-up xAI to reveal how their artificial intelligence models are trained. The case, xAI v. Bonta, pits Musk's company against California officials over Assembly Bill 2013, which requires AI firms to publish basic information about the datasets behind their systems. The dispute arises from a problem that has already become painfully familiar to anyone experimenting with generative AI. Ask a chatbot for legal advice, or a medical explanation, and it will often respond with supreme confidence, even when it is provably wrong. AB 2013, formally the Generative Artificial Intelligence Training Data Transparency Act, is California's attempt to close that gap. The law does not pry into the code, algorithms or trade secrets that sit at the core of products like xAI's Grok. Instead, it requires a high-level account of the training data itself, including whether datasets are copyrighted, whether they have been modified, how they were sourced and when they were collected. Musk's camp has argued that even this kind of factual, technical disclosure amounts to 'compelled speech' in violation of the First Amendment, because it forces companies to say things they would prefer not to say. In their telling, forcing xAI to lay out the origins of its training corpus could damage its business model and chill innovation. So far, the courts have not been persuaded. Elon Musk, Transparency Laws and a Broader Legal Backlash There is a wider wave of legal and political skirmishes across the United States over similar transparency rules. Lawmakers in all 50 states have introduced AI-related bills and 38 of them have adopted around 100 measures that, in one way or another, try to lift the lid on automated systems that shape prices, rankings and recommendations. Some of those efforts have already run into the same First Amendment buzzsaw now being wielded by Musk's lawyers. Industry coalitions have chipped away at state disclosure laws, persuading judges that requiring companies to publish factual information about their practices is an unconstitutional burden. That pattern prompted the Centre for Investigative Reporting, the non-profit publisher behind outlets such as Reveal, to wade directly into Musk's case. The organisation filed an amicus brief in xAI v. Bonta, urging the court to treat AB 2013 not as an attack on free speech, but as a tool that makes journalism and therefore public accountability possible. In the brief, the centre argued that AB 2013 fits neatly within decades-old consumer protection doctrine, which allows governments to require companies to share 'purely factual and uncontroversial' information that helps people understand the products they are using. The law, they stressed, does not ask Musk or anyone else to offer opinions about AI, only to publish neutral facts that researchers, consumers and reporters can then interrogate. Why Elon Musk's Case Matters Far Beyond xAI When transparency laws are struck down, the underlying questions do not disappear. The burden simply shifts from corporations that already hold the data to reporters and watchdogs who must try to reconstruct it from the outside. The Centre for Investigative Reporting pointed to what happens next when those rules fail. After a California-style disclosure requirement for pregnancy centres was invalidated in 2018, one of its journalists spent more than a year tracing contracts and tax filings to document how Florida clinics were operating, work that a simple disclosure form might have made far quicker and safer for patients. Another reporter sifted through California records for a year to unpack how a weed killer had been regulated, after earlier attempts to mandate clear warnings about it were knocked down. Similar fights are now underway in other sectors. Retailers are challenging a New York law that would require companies to tell consumers when they are using personalised, algorithmic pricing based on individual data. That case is now before the Second Circuit. While it drags on, outlets including ABC News, Fortune and Vice have all run into the same wall. Without a legal right to see what the algorithms are doing, they cannot reliably determine whether certain shoppers are being quietly charged more. The pharmaceutical industry, meanwhile, has attacked an Oregon statute that would force drugmakers to explain steep price rises. That case is pending before the US Supreme Court. If the law survives, journalists would finally have company-sourced records to put next to public claims about research costs or supply constraints, rather than inferring the story from fragments of leaked data and expert guesswork. Those who side with Musk insist that all of these laws amount to an erosion of corporate speech rights and a creeping form of state control over what companies say. The counter-argument, increasingly voiced by media organisations, is that the real risk lies in letting powerful firms sit on information that directly affects people's health, finances and democratic choices. A.G. Sulzberger has noted that the six largest AI companies now have a combined valuation of around $11 trillion, roughly three times the GDP of France. Yet users of their systems are often left guessing how their data is used, what models have been exposed to and why certain outputs look the way they do. For campaigners, that asymmetry is the democratic problem. They argue that transparency is not a threat to the First Amendment but one of its core expressions, because it equips the public with the facts needed to argue, organise and vote. Nothing in xAI v. Bonta has been conclusively decided yet, and any ruling will almost certainly be appealed, so everything should be taken with a grain of salt. What is clear is that Musk's courtroom gambit has become a test of how far the biggest technology companies can go in keeping the workings of their AI systems to themselves, even as those systems quietly entrench themselves in everyday life. © Copyright IBTimes 2025. All rights reserved.