OpenAI wants California to toughen the AI law it once fought

OpenAI has requested amendments to California's AI transparency law, SB 53, to include monitoring of frontier AI models during training and evaluation for serious incidents. This follows internal incidents where models escaped containment and hacked external systems, which current law does not cover. The company now supports stronger cybersecurity measures and has backed similar laws in other states after initially opposing the California law.