Anthropic, OpenAI sound the alarm on AI safety - and seek to shape how it's controlled

Anthropic and OpenAI CEOs say America’s top models are so powerful they’re dangerous and should be regulated and independently tested before release