Elon Musk's xAI company has launched an AI robot named Grok, positioned as a "rebellious" robot capable of answering various sharp and sensitive questions. However, tests have revealed that some of its responses inadvertently reference OpenAI's policies, despite Grok's development purpose being to compete with ChatGPT. xAI engineers explain that this is due to unintentional inclusion of ChatGPT outputs in the training data, yet some experts remain skeptical about whether the training involved fine-tuning the ChatGPT model. The article also compares the differences between Grok and ChatGPT, and lists Grok's performance during the tests.