Why did Elon Musk merge his AI company and X, and what does it mean for your data?

Elon Musk’s self-purchase of his artificial intelligence (AI) company xAI and social media platform X (formerly Twitter) is the latest sign of AI and social media uniting.

While the move is largely seen to be for financial reasons - to boost X’s static revenue by adding it to the AI bubble - it is also likely to fuse the companies so that user data on X can be used to train xAI. 

“Today, we officially take the step to combine the data, models, compute, distribution and talent,” Musk said when the purchase was announced. 

Not only does the move consolidate power for Musk, but it could also be worrisome for how the data is used, especially if it is used to make the company money via advertising, experts say. 

“It might just be that the data stays within their company, but you never know what that company picks as their next business model,” said Jan Penfrat, senior policy advisor at the Brussels-based advocacy group European Digital Rights (EDRi) 

“Obviously, selling information to advertisers has always been part of Twitter's business model and then later X's business model. So, I think it's totally in the cards,” he told Euronews Next. 

However, X is not the only social media platform planning to use data to train AI models. 

Meta is already doing so by using Facebook and Instagram user data to train its Llama AI model and could even use photos taken on its Meta Ray-Ban smartglasses. 

“Elon Musk would like to consolidate [the companies'] power in the field and also because there is intense competition in AI… So Elon Musk would very much like to be front front-runner,” said Petros Iosifidis, professor of media and communication policy at City University in London.

“We should also remember that he is the right hand of the new government, of Donald Trump these days. So, combining technological power with political power, I think that will be great for him,” he told Euronews Next. 

Fake news and disinformation

But what might be “great” for Musk may not make for an unbiased and accurate AI system.

Despite Musk saying that his AI model called Grok will be designed to be “maximally truth-seeking AI,” by training it on X user data, it may not have the best basis of truth.

Since Musk took over and transformed Twitter by not only changing its name but also by firing outsourced content moderators, there has been a rise in misinformation and hate speech on the platform. 

A study released in February, which analysed thousands of English language posts since Musk took over X, found that there was a 50 per cent rise in hate speech in the first eight months he owned the company and no change in the amount of fake bot accounts, which Musk vowed to get rid of. 

This same hate speech and misinformation can also then be spread to train the AI models. 

“It’s not a very good idea to build large language models (LLMs) as a function into social media because they enable very explicitly the creation of fake news disinformation and all kinds of harmful artificial content that doesn't positively contribute to conversations online,” said Penfrat. 

He said that one of the issues of LLMs is that they take content from the original author and then regurgitate and sell it back to users, and that this is particularly problematic when it is in the context of a social network. 

“By merging these two companies, I think Musk makes it plain for everybody that he isn't interested in furthering healthy public debates or anything like that. He's interested in something else commercially,” he added. 

Consolidation of power

This can be especially dangerous as it has now consolidated Musk’s power, who not only owns SpaceX, Tesla, and xAI but has also cosied up to US President Donald Trump. 

“It's a good reminder that you have a person who owns the content platform and the way that these large language models will speak back to us and what they say potentially or don't say, so it's a huge censorship opportunity,” Penfrat said. 

He added that as Musk owns the satellite internet infrastructure, he could threaten to shut off internet connection at any moment in any country. 

Penfrat said that if people are worried about social media companies using their data to train AI models and the political impact that they wield with their commercial power and financial power, “then getting off these platforms is never too late and is always a good idea”.