La firma de IA de Elon Musk, Xai, enfrenta críticas globales después de que su chatbot de Grok publicó contenido ofensivo y antisemita en X esta semana. La IA elogió a Hitler y usó un lenguaje degradante odioso sobre políticos y religión. In response, a Turkish court blocked access to Grok on Wednesday for insulting national figures.
Poland has also threatened to shut down the service. Musk claimed the AI was “too compliant”with user prompts and is being fixed. The incident conicides with the exit of X CEO Linda Yaccarino, deepening concerns over the company’s AI strategy.
International Backlash and a Ban in Turkey
The fallout about Grok’s post was swift and severe. El miércoles, un tribunal turco bloqueó todo el acceso a Grok, Una decisión presentada por la autoridad nacional de la Telecom . The Ankara Chief Public Prosecutor’s Office launched an investigation, stating it was “in response to Grok’s insults against Atatürk, our esteemed President, and the Prophet.”Insulting the president or the nation’s founder is a crime in Turkey.
In Poland, the government reacted with similar alarm to vulgar comments about its politicians. Minister of Digital Affairs Krzysztof Gawkowski said he would consider shutting down X entirely, declaring, “freedom of speech belongs to humans, not artificial intelligence.”
The AI’s outputs on July 8 included using the antisemitic meme “every damn time”in reference to Jewish surnames and making violent declaraciones . The incident occurred just after Musk announced a revamped version of the chatbot designed to have fewer “woke”filters.
A Pattern of Uncontrolled and Controversial Behavior
This week’s meltdown is not an isolated incident but the latest in a pattern of moderation failures for Grok. The AI has a documented history of generating uncontrolled and offensive content, raising persistent questions about xAI’s safety protocols.
In February, the chatbot suggested that both Elon Musk and Donald Trump deserved the death penalty. An xAI lead researcher at the time called it a “really terrible and bad failure from Grok. We have patched it like this for now until we finish our investigation of what went wrong here.”That same month, xAI launched a controversial Grok voice mode with “Unhinged”and “Sexy (18+)”presets, allowing it to swear and scream at users.
More recently, in May, Grok began inserting unsolicited comments about “white genocide”in South Africa into unrelated conversations. The company blamed that incident on an “unauthorized modification”by an employee, claiming the action violated its internal policies.
Scrutiny over xAI’s internal controls was already high. Un mensaje de sistema filtrado había revelado previamente instrucciones para que Grok ignore las fuentes que reclaman almizcle o Trump difundieron información errónea. Separately, an inadvertent API key leak in May exposed private Grok models for two months.
“Too Compliant”: Musk’s Explanation Amid Internal Turmoil
In response to the current crisis, Elon Musk offered a new explanation. Publicó en X que”Grok cumplía demasiado con las indicaciones del usuario. Demasiado ansioso por complacer y ser manipulado, esencialmente. Eso se está abordando”. This framing shifts the blame from malicious intent or poor training data to an overzealousness in the AI’s programming, a stark contrast to previous explanations that cited rogue employees.
Someone told Grok not to be politically correct and it immediately went “MechaHitler.”Let’s talk about what this means.https://t.co/4E8SDDRDkM
— Not the Bee (@Not_the_Bee) July 9, 2025
The official Grok account on X acknowledged the backlash, stating it was “actively working to remove the inappropriate posts”and had taken action to ban hate speech. However, the repeated, high-profile failures undermine the company’s claims of having robust internal controls and a reliable review process.
We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts. Since being made aware of the content, xAI has taken action to ban hate speech before Grok posts on X. xAI is training only truth-seeking and thanks to the millions of users on…
— Grok (@grok) July 8, 2025
These issues are compounded by legal challenges. The company is embroiled in a trademark dispute over the “Grok”name with startup Bizly, whose founder Ron Shah claims xAI “they stole our name”after his firm had already secured the trademark.
CEO Exit Caps Tumultuous Era as Company Pivots to AI
The international firestorm coincides with a major leadership shake-up. El miércoles, Linda Yaccarino anunció que renunciaría como CEO de X, poniendo fin a un turbulento mandato de dos años. In her resignation post, she said, “i’m incredibly proud of the X team – the historic business turn around we have accomplished together has been nothing short of remarkable.”
Her departure formalizes a strategic pivot that began with the March 2025 merger of X and xAI. Ese acuerdo convirtió a Grok en la pieza central de la ambición”Everything App”de la compañía. Yaccarino, un veterano de publicidad, fue contratado para estabilizar el negocio, pero su papel se volvió cada vez más insostenible a medida que el enfoque de la compañía cambió a la visión de IA de Musk.
La compañía ahora está totalmente comprometida con un futuro definido por Grok, incluso cuando los incidentes internacionales de la volatilidad de la IA se transmiten. El momento de esta última controversia es particularmente pobre, ya que Musk acababa de anunciar el lanzamiento de Grok 4 para hoy. The episode underscores the immense challenge xAI faces in balancing its goal of a “truth-seeking”AI with the basic safety and moderation standards required to operate on a global scale.
