OpenAI, Anthropic warn powerful AI could pose serious risks . But who gets to set the safety rules?

OpenAI and Anthropic are calling for independent testing and stronger safeguards for advanced AI models as recent agent incidents sharpen debate ...