ChatGPT cites Grokipedia: OpenAI finds a new source of truth

The world of artificial intelligence is small, but we didn't expect it to shrink this much this fast. ChatGPT users have found something odd: the chatbot has started pulling data from Grokipedia. In case you missed the launch, Grokipedia is an ambitious xAI project from Elon Musk that pitches itself as a knowledge base without censorship or a left-wing agenda. The irony is off the charts: Sam Altman's company, which Musk regularly accuses of being too progressive and too closed, now leans on Musk's own creation to shape its answers.

To understand how we got here, it helps to recall the history of the conflict. Musk was at the origin of OpenAI, then left, slamming the door on his way out, and launched his own xAI with an anti-woke slant. Grokipedia is his answer to Wikipedia, which Musk considers too biased. It is a huge body of data generated and moderated by xAI's algorithms. And now that data is surfacing in ChatGPT. Why now? The answer is simple and worrying at once: OpenAI badly needs fresh data, and the company is not especially picky about how it gets it.

OpenAI's web crawlers comb the internet around the clock. When xAI released Grokipedia into open access, it automatically became fair game for search bots. It seems Altman's engineers did not put a filter on Musk's domains, or the ranking algorithms judged Grokipedia's content sufficiently relevant. The result is a curious cocktail. ChatGPT, which usually tries to steer clear of sharp edges, suddenly serves up facts or interpretations typical of Grok. This is not just a technical curiosity; it is a sign of an approaching data crisis across the whole industry.

We are close to the point where high-quality human content on the web simply runs out. Developers of large language models are starting to feed on content produced by other models. In professional circles this is called model degradation or digital inbreeding. If ChatGPT learns from Grokipedia's data while Grok keeps training on ChatGPT's answers, we risk ending up with a closed echo chamber. In such a system, errors and biases multiply exponentially, gradually pulling AI away from reality.

For Elon Musk, the situation looks like a double-edged win. On one hand, his knowledge base is considered weighty enough to be cited by the market leader. On the other, OpenAI is using for free a resource on which xAI spent millions of dollars and enormous computing power. It recalls Reddit and Twitter, which closed their APIs so that AI giants would not take their content for training. It is quite likely that Musk's next step will be to block OpenAI's bots from Grokipedia, or another high-profile lawsuit over intellectual property infringement.

The borrowing of data between neural networks raises an important ethical question. If one AI starts citing another AI without a direct link to the original source, how will we verify the accuracy of the information? For now, we are watching the two main rivals in the industry unwittingly merge into a single information field. That could lead either to a universal superintelligence or to a complete collapse of meaning, where one neural network simply retells the hallucinations of another.

Bottom line: Will this case become grounds for a new copyright war between Musk and Altman, or have we officially entered an era in which AIs train each other until they lose all connection to the human original source?