Tech
screenshot •
Grok has influenced Trump’s decisions
The US president is reported to have consulted Elon Musk’s artificial intelligence before the attack in Venezuela. The Democrats are calling for the conversations to be preserved: the fear is that the chatbot may have influenced his decisions

Donald Trump looks at his smartphone before boarding Air Force One (GettyImages)
In late October, the U.S. House of Representatives’ Oversight and Judiciary Committees asked Donald Trump to preserve his conversations with the Grok chatbot to understand what they were about. The development was soon overshadowed by the mid-term elections and the Democrats’ victory, but it is worth revisiting because Trump is not exactly your average user, and Grok is a service developed by xAI, a company owned by Elon Musk. In short, all the ingredients are there for a cyberpunk noir.
The story officially began in early October, when the weekly magazine Time published a lengthy article on the last two years of the Trump administration, focusing in particular on his efforts to refurbish the White House (whose interiors are now covered in gold, in the style of Trump Tower) and to build the new reception hall.
However, it was one paragraph in particular that caught many people’s attention: it recounts how, last December, Elon Musk returned to Washington after months of a war of words with Trump. Musk brought a gift with him: Grok, the chatbot developed by one of his companies, xAI.
According to sources, Trump began experimenting with the platform, submitting a series of questions to the artificial intelligence about his presidency and his political legacy. Although the former president is not widely regarded as a tech enthusiast (he is, after all, an 80-year-old man), operating a chatbot is well within the capabilities of a seasoned Twitter power user: one simply types inquiries as in any standard chat, and Grok handles the rest.
A bit of background. Back in December, the United States was already considering striking Venezuela to oust its leader, Nicolás Maduro. At the time, these attacks mainly targeted vessels accused of drug smuggling into the United States, but Trump wanted more. Undecided, he asked Grok how a military attack would be received by the Venezuelan people; the chatbot replied that Maduro was an unpopular and authoritarian leader, and that therefore a large part of the population would welcome it. Who knows, perhaps they would have thanked Trump for freeing them from his yoke.
During the night of 2–3 January, Maduro was captured by the military, arrested and taken to the United States. The news caused consternation around the world but also great joy amongst Venezuelans. In short, Grok had been right, and Trump emerged from the whole affair convinced of one thing: that AI was ‘ingenious’.
The entire report rests on a single source cited by Time—anonymously—and should therefore be treated with caution. Nevertheless, if Trump found Grok useful on that occasion, it is reasonable to assume that he may continue to consult it. It is therefore worth asking what the U.S. President discusses with his chatbot and what sort of confidential information he shares with Grok. But above all, what on earth does the chatbot actually reply?
This is what the Democrats intend to find out by requesting that Trump’s chatlog with Grok – that is, the exchange between the user and the AI – be preserved. And it is likely that this is not a problem limited to this administration: with the spread of these tools, more and more politicians are using (or will use) AI in their work, and specific rules may well be needed.
In the meantime, however, Grok is the least of the Democrats’ worries as they prepare to take control of the House and the Senate, and have already expressed their concern about the so-called ‘paper trail’. That is to say, the paper trail – all the internal documents, communications, plans, and orders issued and received by Trump and his administration – which will be crucial to the investigations the Democrats wish to carry out – and which they may not necessarily find.
There is a very real risk that the document shredders will be activated and everything will disappear, even though it is well known that Trump and his entourage prefer Signal, the ultra-secure messaging app, which has a feature where messages self-destruct. Yet the case of Grok, if confirmed, is worrying in a different, new way: a chatbot is, in fact, a powerful technology; it can advise, persuade, charm and, why not, amaze. And the risk is that it may have influenced – at least in part – Trump’s decisions over the past year.