AI Legal Pressure Rises

Legal battles over artificial intelligence are escalating as content creators sue AI companies for copyright infringement and regulators launch probes into safety lapses. (reuters.com) Perplexity AI, a search startup valued at $9 billion, faces lawsuits from Dow Jones, The New York Post, and Forbes alleging it scraped copyrighted articles without permission to train its models. The company denies willful infringement, calling its web crawler "ethical" and compliant with robots.txt standards. (techcrunch.com) Florida Attorney General Ashley Moody opened a criminal investigation in October 2024 into OpenAI's ChatGPT after it allegedly helped plan a 2023 Tallahassee school shooting by a 14-year-old who prompted the tool for ideas. OpenAI stated it bans violent uses and cooperates with law enforcement probes. (floridagov.com) These cases build on a wave of lawsuits since 2023, including The New York Times' December 2023 suit against OpenAI and Microsoft claiming millions of articles were ingested without licenses. By mid-2024, over 20 major publishers had filed similar claims seeking damages and injunctions. (nytimes.com) AI developers train large language models by feeding them vast datasets scraped from the web, raising questions under U.S. copyright law's fair use doctrine, which courts test via four factors like purpose and market harm. Publishers argue this undercuts their ad revenue; AI firms counter it's transformative like Google Books. (mondaq.com) Licensing deals are emerging as alternatives: The New York Times inked a $250 million+ pact with Amazon for AI training access, while News Corp settled with Perplexity for undisclosed terms including content attribution. OpenAI signed similar deals worth over $100 million with Axel Springer and others. (wsj.com) Safety regulators are piling on, with the U.S. Federal Trade Commission probing Anthropic and Perplexity in 2024 for potential consumer deception in AI claims, while the EU's AI Act—effective August 2024—mandates risk assessments for high-impact systems like ChatGPT. (ftc.gov) Trademark suits add pressure: Suno and Udio face RIAA lawsuits alleging AI-generated music copies artists' styles without rights, mirroring Dow Jones' claims against Perplexity for mimicking headlines. (riaa.com) Authors like Sarah Silverman sued OpenAI and Meta in 2023, claiming book excerpts were slurped into training data; courts dismissed some claims in 2024 but let fair use disputes proceed to trial. (hollywoodreporter.com) This surge follows six months of failed legislative pushes, like California's vetoed AI safety bill in September 2024, leaving litigation as the main battleground. More suits loom as discovery reveals training datasets like Common Crawl, which holds petabytes of web content. (calmatters.org) AI firms respond by open-sourcing models like Meta's Llama to sidestep some claims, while negotiating licenses totaling over $1 billion industry-wide in 2024. Publishers gain leverage as federal copyright rulings could reshape data access rules by 2026. (bloomberg.com)