翻訳待ち：Elon Musk is suing California over an AI law. Here's why transparency matters

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

Ask an AI chatbot a question, and it answers with total confidence — even when it’s wrong. Chatbots have invented court cases, medical dosages and statements no one ever said. The companies that built these tools know exactly what data produced that confidence, but you don’t. That gap is the fight now playing out in courtrooms nationwide and in California. Elon Musk is at its center. This year his company, xAI, sued California over a law demanding that artificial intelligence companies disclose how they train their models. Musk’s company claims the law violates its First Amendment right to withhold speech by compelling transparency. But California’s law is not an outlier. Assembly Bill 2013, the Generative Artificial Intelligence Training Data Transparency Act, is part of a nationwide wave of legislation — all 50 states introduced AI-related bills, and 38 adopted roughly 100 measures — aimed at prying open the black box of technology now embedded in daily life. It requires AI companies to publish a high-level summary of their training datasets: Is it copyrighted? Was it modified? When was it collected? None of the law’s roughly dozen factual categories touches on proprietary methods or requires opinions from the companies; it simply gives researchers, users and journalists enough facts to judge these datasets for themselves. Anthropic and OpenAI have already complied with AB2013’s requirement to provide short public summaries. xAI sued instead, arguing the law unconstitutionally compelled it to divulge proprietary information. The government argues there are no First Amendment violations when the government interest in public disclosure is reasonable. But the company frames the public’s interest in the data as mere “curiosity.” The mere curiosity argument didn’t persuade the U.S. District Court. Judge Jesus Bernal wrote, correctly, that it “strains credulity” to suggest no consumer could usefully evaluate information about an AI model’s training data. The judge relied on a long line of First Amendment cases that have elevated the consumer’s right to useful information for a host of products. The consumer’s right to know is especially urgent with AI, as its hallucinations creep into everyday life. Being transparent about how data was trained isn’t a nicety; it’s the information a user needs to think critically about how they use this technology. And while the average person might not read disclosures required under the law, a journalist will. Like laws governing food labeling, toxic chemical releases and consumer privacy, AB2013 is essential for reporters researching corporate conduct. Unfortunately, a wave of industry-led “compelled speech” challenges, like xAI’s, has struck down an increasing number of transparency laws, making reporting harder, ultimately disempowering the public. That is why the Center for Investigative Reporting filed an amicus brief in the case, xAI v. Bonta. The center argued that compelled speech laws are important for journalists to do their work and that AB2013, in particular, complies with the First Amendment because it demands high-level information important to the consumer, protected by decades-old doctrine safeguarding consumers’ right to know. When transparency laws are overturned, the burden of finding the truth doesn’t disappear — it shifts from a company that can readily access the information to a reporter who must attempt to uncover the facts, in a lengthy, often years-long investigation. After one disclosure law was struck down regarding pregnancy centers in 2018, a Center for Investigative Reporting journalist spent over a year piecing together contracts and tax filings to expose lax oversight of pregnancy centers in Florida for a 2023 story. Another journalist spent a year reconstructing California records to understand aspects of a weed killer product that had previously challenged a disclosure law about it. These delays risk the public’s health and safety. The same pattern is playing out around the country. Last year, retailers challenged a New York law requiring companies to disclose when they adjust prices based on algorithmic pricing that considers a shopper’s personal data. Currently, the case sits before the Second Circuit. In the meantime, ABC News, Fortune and Vice have all hit the same wall: Without mandated disclosure, there is no reliable way to confirm whether prices were raised based on what the company’s algorithm knew about its customers. Drug pricing shows an identical dynamic: the pharmaceutical industry has challenged an Oregon law requiring drugmakers to explain price increases, a case pending before the U.S. Supreme Court. If it survives, the law would give reporters a company-sourced record to check against public claims, instead of forcing them to reverse-engineer pricing logic secondhand, as outlets like Kaiser Health News have had to do. Companies invoking the First Amendment argue that disclosure laws like these threaten their rights and business models by compelling them to speak. But as New York Times Publisher A.G. Sulzberger recently noted, the top six AI companies are valued at $11 trillion, three times France’s gross domestic product, and still often leave their users in the dark about their data. Holding corporations like them accountable requires laws like AB2013 because they won’t voluntarily disclose information that impacts consumer choices — and the public has a right to it, a principle the Supreme Court recognized 50 years ago. Transparency is not a free speech violation. Instead, it is a First Amendment value that protects the audience, ensuring the public is informed about the systems and critical consumer choices shaping their lives. We cannot allow “compelled speech” arguments to dilute the value of truthful facts about technology we rely on for critical aspects of our lives. The health of our democracy depends on it. Victoria Baranetsky is the general counsel of the Center for Investigative Reporting, the nation’s oldest nonprofit investigative newsroom. Brooke Henderson is the First Amendment Fellow at CIR.