No way to establish spending limits

Let me get this straight: there’s no way to control the cost of inference? Does this mean that if my API key gets compromised, I could end up paying thousands of dollars because there’s no way to establish spending limits and no system in place to warn about usage?