Regulators said the company’s lack of controls had led to the widespread use of deepfakes created with the chatbot Grok.