Jonathan Durand Folco, Associate professor, Université Saint-Paul / Saint Paul University
Elon Musk’s update of his chatbot Grok went seriously awry recently when it started spouting antisemitic comments and presented itself as “Mecha-Hitler.” Less than two weeks later, xAI, the company that developed Grok, secured a US$200 million contract with the Pentagon in the United States.
Should we be nervous about the unregulated artificial intelligence feeding the U.S. military machine?
Among the many chatbots on the market, Grok stands out for claiming to be free from the constraints of “political correctness.” The AI tool, one of Musk’s subsidiaries, aims to surpass the capabilities of its competitors (including ChatGPT, Claude, Gemini and Copilot) while taking an “anti-woke” stance, free from “censorship” or overly rigid moral rules.
In April 2023, the billionaire entrepreneur launched the TruthGPT project that aims to create “maximum truth-seeking AI that tries to understand the nature of the universe” while offering an alternative to ChatGPT’s “liberal bias.”
It should be noted that Musk, who cofounded OpenAI in 2015 but left in 2018 after failing to become its CEO, has allegedly been seeking revenge ever since on his rival Sam Altman became a superstar in the digital industry after the global success of ChatGPT.
My research focuses on the changing face of capitalism, the social impacts of AI and contemporary authoritarianism. I recently published Le capital algorithmique (Algorithmic Capital) with my colleague, political theorist Jonathan Martineau.
Grok, a distorting mirror of its creator
Like its creator, Grok is characterized by dark humour and a rebellious streak, openly displaying contempt for social norms and taboos. The first major problem is its propensity to spread fake news, especially through its integration into the social media platform X, also owned by Musk. Like other generative AIs, Grok suffers from bias, hallucinations and inaccuracies.
À lire aussi : What are AI hallucinations? Why AIs sometimes make things up
But the absence of ethical and factual guidelines in this “censorship-free” conversational robot can produce absurd results, such as realistic images of Musk and Mickey Mouse killing children or promoting the narrative of white genocide in South Africa.
The idea of anti-woke AI also dramatically amplifies misinformation in the name of freedom of expression, with no attached responsibility to correct it.
When the far right flirts with high tech
A second problem arises when the richest man in the world rallies behind Donald Trump’s second administration.
Musk’s high-profile involvement in Trump’s presidential campaign, his adherence to the reactionary ideology of Dark MAGA, his apparent Nazi salute on Jan. 20 and his support for the Alternative for Germany party (AfD) all testify to his shift to the far right of the political spectrum.
Still convinced his AI was too woke and heavily influenced by traditional media, Musk announced an update to Grok 4 in early July. That’s when the bot quickly started generating antisemitic, neo-Nazi and conspiracy-theory-laden speech, referring to itself as “Mecha-Hitler” (the start-up xAI has since apologized). This is disturbing but hardly surprising: the creature often reflects its creator.
When technological power infiltrates the state
A third problem arises when the interests of the private and public sectors converge to an extent that can be described as collusion or even corruption. Musk was hired as a special employee of the Department of Government Efficiency (DOGE) to make massive cuts in the federal administration. But his own companies — Tesla, SpaceX and Neuralink — have racked up more than US$2.37 billion in federal contracts.
Some commentators have described DOGE’s use of AI and “move fast and break things” logic as an “AI coup.”
Ironically, back in March 2023, Musk signed the open letter Pause Giant AI Experiments, which was supported by hundreds of public figures. At the time, members of the research community, industry, public sector and civil society were united in calling for a moratorium on the development of generative AI because of the many risks it posed: misinformation, surveillance, job losses, environmental impacts and threats to democracy. It read:
“Should we let machines flood our information channels with propaganda and untruth? […] Such decisions must not be delegated to unelected tech leaders. Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable.”
In 2025, the opposite scenario is prevailing: AI is spreading false content on a massive scale, major decisions are being entrusted to unelected private entities and artificial intelligence is becoming a lever for economic, political and military domination.
This broad convergence between the Silicon Valley oligarchy and the populist MAGA movement is sometimes referred to as techno-fascism. This is not a simple matter of complicity between Musk and Trump, which recently came to a spectacular end with an exchange of online insults. It is about the structural integration of algorithmic technologies into an authoritarian project.
This project aims to dismantle the rule of law and establish a police state, inspired in part by Hungarian President Viktor Orban and his “illiberal democracy.”
From Mecha-Hitler to helping the army
A final example of the integration of AI in the service of techno-fascism is the contract signed between xAI and the U.S. Department of Defense in mid-July.
The military wants to use Grok to provide “support capabilities to combatants,” and help “maintain a strategic advantage over their adversaries.” Even though Musk has officially left DOGE and broken ties with Trump, his “Mecha-Hitler” AI will be directly connected to American armed forces.
With Trump’s total deregulation of the digital sector, the adoption of the Big Beautiful Bill that increases the budget of the U.S. Immigration and Customs Enforcement agency, and the US$500 billion Stargate project, we are on the threshold of a techno-fascist era.
Jonathan Durand Folco ne travaille pas, ne conseille pas, ne possède pas de parts, ne reçoit pas de fonds d'une organisation qui pourrait tirer profit de cet article, et n'a déclaré aucune autre affiliation que son organisme de recherche.
