Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages

Would this be enough to keep an AI model from going out of control?