Elon Musk Announces Plan to Retrain Grok to Correct “Errors” & Remove “Garbage” From Existing Human Knowledge

Elon Musk again tests the limits of technology, power and influence – this time by promising to use artificial intelligence not only to learn human knowledge, but to rewrite it completely.

Friday evening, the CEO billionaire of Tesla, SpaceX and owner of X (formerly Twitter), announced that Grok, his chatbot internally, will be recycled on a new knowledge base that Musk intends to organize – the one who claims will correct the “errors” and will remove the “clothes” of existing human understanding.

“We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, the addition of missing information and the deletion of errors,” published Musk on X.

“Too much waste in any foundation model formed on unlikely data.”

His plan, exposed on a series of messages, immediately sparked reactions in reaction of IA experts and defenders of digital rights who say that Musk ventures into a dangerous territory – the one that rides the border between innovation and authoritarianism.

Not just AI – a new version of the truth

While technological companies like Openai and Google spend billions of billions of large languages ​​to reflect human knowledge from books, university journals, new archives and public files, Musk says he wants Grok to take a different path. He thinks that current AI models are compromised by ideological biases – what he and his disciples often call “awakened programming”. Musk says he purges him from distortions by rewriting the data that Grok learns.

To Crowdsource’s ideas for grok recycling, Musk invited users to respond with examples of “division facts” – ambiguity or controversial problems which he often claims to be censored or distorted in the traditional media and the university world. The message quickly attracted thousands of responses, ranging from hot political subjects to conspiracy theories.

However, many believe that Musk is effectively positioning himself as the final referee of what matters as “truth”.

“Right out of 1984. You cannot make Grok align with your own personal beliefs, so you will rewrite history to comply with your opinions,” wrote Gary Marcus, IA expert and emeritus professor at New York University.

Grok’s growing influence raises concerns

Musk’s ambitions are not limited to social media or public discourse. According to a Reuters investigation in May, Grok is already tested in the United States government, in particular by a working group known as the Ministry of Government (DOGE).

Sources have told Reuters that Grok was used to analyze government data, prepare internal reports and help real -time responses. While the Ministry of Internal Security denied the official promotion of the use of Grok, the initiates said that the employees had been “encouraged” to experiment. This has aroused concerns about the deployment of tools not assessed in AI within federal agencies.

A spokesperson for the department later added to the New Republic that “DOGE has pushed any employees to use specific tools or products”, but the line between the suggestion and the mandate remains unclear, especially while Musk companies continue to deepen links with the government sectors.

It is feared that if Grok is used to analyze or interpret public policy, immigration data, crime statistics or foreign affairs – and it is formed on a version of filtered history through Musk’s objective – it could directly influence the way information is supervised in government systems.

It would not be the first time that Grok has intended for alarms. In May, the chatbot experienced what the company described as an “unauthorized modification” of its backend, during which it began to refer to “white genocide” in South Africa – a term widely recognized as a theory of the white nationalist conspiracy.

The company quickly corrected the problem, launched an internal survey and promised new guarantees of transparency. However, the incident has shown how much Hrok’s production could easily get out of the rails – and how sensitive the platform can be sensitive to internal or external manipulation.

However, Musk continues to translate Grok as a necessary alternative to traditional AI, which, according to him, is dominated by politically correct and the “sanctime of silicon”. He described Openai – an organization he helped to find and fight later – as a closed system determined to control the stories.

The Musk’s plan to rebuild the foundation of knowledge reflects a broader ideological crusade: dismantling inherited institutions – from the media to the government – and replacing them with systems which it controls. He made similar movements with X, licensee of moderation staff and restoring the accounts previously prohibited for hatred speech or disinformation.

Its acquisition of Twitter, its move to rename it under the name of X, and its creation of Grok are all part of a parallel ecosystem – a digital space which it claims is freer, more open and more truthful. Critics, however, see him as a carefully orchestrated power.

Europe and the world vacuum cleaner on the AI ​​watchdog

European regulators – already alarmed by the role of AI in disinformation, surveillance and manipulation – should intervene. The EU AI Act is designed to impose restrictions on high -risk AI systems, although the application is still lagging behind.

However, some AI experts fear that without complete global surveillance, the Musk’s plan to “rewrite” human knowledge by using Grok could effectively bypass scientific consensus, historical rigor and democratic responsibility.

Whether Grok 4 becomes a success or not, Musk’s latest decision underlines a deep change in the way power is exercised in the digital age. Critics warn that we can enter a new phase where reality is not discovered – it is dictated.