Litigation Filling AI Gaps
- A recent episode examined whether lawsuits, such as the OpenAI case, are establishing AI accountability where statutes lag. - The discussion covered how litigation can shape disclosure norms, training-data practices, and duty-of-care expectations for firms. - If courts set compliance norms, agencies and contractors may face shifting legal exposures that complicate procurement and oversight. (youtube.com)
Courts are starting to set some of the practical rules for artificial intelligence while Congress and statehouses are still writing them. The OpenAI cases in New York have become a test of how judges may define training-data limits, disclosure duties, and user-privacy boundaries. (courtlistener.com) The New York Times sued OpenAI and Microsoft in the Southern District of New York on December 27, 2023, alleging copyright infringement tied to training and output. Authors Guild filed a separate case against OpenAI on September 19, 2023, and those disputes now sit inside a broader multidistrict copyright proceeding. (courtlistener.com 1) (courtlistener.com 2) (courtlistener.com 3) The cases are not only about money damages. OpenAI said in 2025 that The Times and other plaintiffs sought retention of consumer ChatGPT and application programming interface data beyond the company’s normal deletion timelines, turning discovery fights into a second dispute over privacy and recordkeeping. (openai.com) That is how litigation fills gaps left by slow lawmaking. A lawsuit forces companies to explain what data they collected, how they used it, what safeguards they had, and what they told customers, even when no AI-specific statute spells out each step. (ftc.gov) (oag.ca.gov) Federal agencies are also moving by guidance rather than a single AI law. California Attorney General Rob Bonta said on January 13, 2025 that existing state consumer-protection, civil-rights, competition, and privacy laws already apply to companies that develop, sell, or use AI. (oag.ca.gov) The U.S. Copyright Office has taken the same incremental path. It published Part 2 of its AI report on January 29, 2025 and released a pre-publication Part 3 on May 9, 2025 focused on generative AI training, licensing, and possible liability for use of copyrighted works. (copyright.gov 1) (copyright.gov 2) Outside copyright, the Federal Trade Commission has signaled that old rules still bite. The agency said in January 2024 that model providers can face liability if they break privacy promises, including promises not to use customer data to train or update models, and in September 2024 it announced five AI-related enforcement actions under what it called Operation AI Comply. (ftc.gov 1) (ftc.gov 2) Companies are responding by buying licenses as well as fighting in court. OpenAI has signed content deals with Axel Springer, the Financial Times, Axios, and Guardian Media Group, creating a market signal that some publishers expect payment and attribution rather than relying only on fair-use defenses. (openai.com) (openai.com) (openai.com) (openai.com) Government buyers may feel the effects next. The Government Accountability Office said in a 2026 report that federal agencies more than doubled reported AI use from 2023 to 2024, and the General Services Administration is weighing draft contract terms that would require disclosures about AI systems used in contract performance. (gao.gov) (gibsondunn.com) If judges keep defining what counts as consent, copying, retention, or reasonable safeguards, those rulings will travel beyond Silicon Valley. They will show up in vendor questionnaires, procurement clauses, insurance underwriting, and the next round of AI cases already building on the first. (courtlistener.com) (gao.gov)
