Related videos:
The official media outlet Cubadebate claimed that the artificial intelligence assistant Grok, developed by xAI, the company of Elon Musk, provides "ideologically biased" responses when discussing topics related to Cuba.
In the article “When AI Takes a Stand: What Grok Says About Cuba”, published on December 9 by the Cubadebate Media Observatory, 30 responses generated by the chatbot on the social network X (formerly Twitter) were analyzed.
According to the text, the study classified Grok's responses into three categories: R (right-wing or anti-communist), C (mainstream human rights criticism), and N (neutral). The result was that “22 out of the 30 responses were classified as R (73.3%), 6 as C (20%), and only 2 could be considered N (6.7%).”
The Observatory concludes that "the promise of neutrality is far from being fulfilled," and that Grok "tends to explain the Cuban reality from liberal-conservative and anti-communist frameworks, even when the questions are neutral or critical of the right."
The analysis cites statements from the chatbot, such as "the embargo does not stop Cuba from negotiating with the rest of the world" and "the sanctions are not the main cause of the failure of Cuba and Venezuela; the problem lies in central planning and the destruction of incentives."
According to Cubadebate, the model "organizes information about Cuba within a liberal-conservative discursive framework" and, by doing so, "can help reinforce certain narratives and silence others."
The post by Cubadebate on its Facebook page triggered a wave of responses. Most comments either supported Grok's claims or criticized the article from the media outlet.
"AI really knows what it's saying," wrote one of the early users. Another added, "Grok tells the truth; that's why Elon Musk programmed it as the AI most committed to free speech." Some reacted with irony: "Oh, did the AI make a mistake too?" or "Real Cubans say no more communism, democracy and freedom."
Other comments pointed directly at Cubadebate with messages like "Cubadebate, listen to your people too," "They could observe the state of opinion of the people of Cuba, they don’t see that," and "You don’t need to be an AI to know all of this."
Several users argued that the chatbot simply reflected Cuban reality: "The AI doesn't take sides; it states what aligns best with common sense. Whether it's convenient for you is another matter," wrote an internet user. Another pointed out: "Grok is summarizing what people think about Cuba and its system of government, so draw your own conclusions." Along the same lines, someone commented: "Even the AI has noticed the exaggeration and whining about the blockade."
Some took it mockingly: "Well, now we've found a new culprit for this disaster: AI," "Wow, they've discovered the wheel," "AI definitely doesn't make mistakes with you," and "Now it's all Grok's fault."
There were also more extensive criticisms: “The government of Cuba does not practice a socialist system. It doesn’t even have a system. The government is hostile and manipulative, justifying its mistreatment of the people for decades with the blockade.” Another user wrote: “The Media Observatory of Cubadebate conducted a study and found that Grok has an anti-communist bias. A global scandal! If the regime hates it, it’s because it speaks painful truths.”
Several comments agreed that the chatbot's responses were not ideological, but descriptive: "It states what is true, what socialism has demonstrated for years," "Cuba is in its current state due to poor administration, not because of the blockade," and "Now even AI doesn't support them; even the algorithms have noticed."
Some mocked the media's tendency to find scapegoats: "The touch, the blockade, and now Grok... there's always a new culprit" and "If State Security could imprison AI, they would." In a more critical tone, another user wrote: "You have already lost the battle of ideas for being liars. AI doesn't make mistakes; you do."
Only a few comments attempted to provide nuance to the debate, suggesting that "the causes of the problems are multifactorial" or that "AI can reflect dominant models." However, the overwhelming majority expressed criticism towards Cubadebate and support for the responses of the chatbot.
Grok is the artificial intelligence assistant created by xAI, a company founded by Elon Musk, and integrated into the social network X (formerly Twitter). Musk has described it as a model designed to respond "with humor and without censorship."
The Cubadebate article mentions academic studies on political biases in language models, but it uses the Cuban case to warn about the ideological impact that, according to the outlet, these tools can have on international perceptions of Cuba.
The controversy between Cubadebate and Grok illustrates how artificial intelligence has entered the realm of ideological dispute. While the official media accuses the model of reproducing “liberal-conservative discourses,” the reaction on social media reflects that many Cubans perceive the chatbot's responses as merely a reflection of their everyday reality.
Filed under: