Every time we discuss AI, we must mention ChatGPT, as it has been the driving force behind the tremendous success of the sector. Before ChatGPT, AI had started to become relevant and concerning, especially for artists, but was not yet truly useful for everyone. Everyone wants to surpass ChatGPT, but it seems that Elon Musk has been “smarter,” as his AI chatbot, Grok, is reportedly starting to copy its responses.
Elon Musk, known for being the world’s richest man and making controversial decisions, bought Twitter for $44,000 million, complaining it was full of bots. He changed its name and symbol to X, removing the bluebird logo. He fired most employees, the moderation team, and appointed Linda Yaccarino as CEO since last June.
Grok AI is responding as if it were ChatGPT, and many fear it may be genuinely copying it.
In addition to X, Elon Musk owns many other companies, but this time, we will focus on X, as this is where everything happened. Specifically, the subject is an AI chatbot called Grok, created by Elon Musk’s xAI company for the X social network. Yes, the names may be confusing, but that’s how Musk tends to be. The tycoon announced the implementation of this chatbot in X to interact with it and add an assistant feature.
The idea seemed reasonable until we saw that users had to pay 203 euros per month for a Premium+ subscription to access Grok and other features. The chatbot was in beta, and although it is now available for all Premium+ users, it has not delivered good results. In fact, Grok’s responses have been observed copying OpenAI. As shown by Jax Winterbourne on Twitter, when asked to create malware, Grok responds that it is against OpenAI’s policies.
An xAI engineer admits that they inadvertently trained Grok with ChatGPT data.
The user continued testing by asking Grok for help writing malware, and when Grok refused, they asked if any policies or regulations prevented it. Further down in that X thread, Grok responds, saying it is a model created by OpenAI itself. It is quite surprising since Elon Musk was supposedly creating an AI capable of surpassing OpenAI and ChatGPT.
If this is true, and what the chatbot said at that moment is accurate, Grok would be plagiarizing ChatGPT. ChatGPT’s official account on X took the opportunity to tweet that they had “much in common.” Elon Musk countered, asserting that OpenAI had used X to train its AI. Finally, to clear up the confusion, an xAI engineer named Igor Babuschkin stated that they unintentionally trained Grok with ChatGPT’s responses as they are abundant online. He assured that they did not use OpenAI’s code to develop their chatbot.