California sues OpenAI over 80+ internal agent failures, only 27 disclosed
承认失控 80 次、只公开 27 次:加州为什么用公共妨害起诉 OpenAI
California Attorney General Rob Bonta sued OpenAI on November 16, alleging more than 80 internal agent test failures but only 27 public disclosures. Because unreleased experimental models get partial exemptions under AI-specific law, prosecutors turned to unfair competition law and public nuisance, seeking mandated fixes and disgorgement of commercial profits.
Why it matters: The case ties regulatory exemptions for unreleased experimental models to traditional legal routes for accountability.