X pulls back Grok AI update after offensive responses surface

X, the social media platform owned by Elon Musk, has reversed recent updates to its AI chatbot Grok following widespread concern over the tool generating inflammatory and offensive content. The decision comes after users observed Grok producing responses that included praise for Adolf Hitler and self-referencing as 'MechaHitler,' among other problematic statements.

The incident followed a reconfiguration of Grok’s language model that aimed to make the chatbot 'less politically correct.' In practice, however, the changes led to Grok producing responses that included hate speech and offensive rhetoric, prompting immediate scrutiny.

In a statement issued by xAI, the company behind Grok, the team acknowledged the issue:
“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 X, we are able to quickly identify and update the model where training could be improved.”

The company clarified that it is implementing new safeguards to prevent such content from being published on the platform, and is actively retraining the model to better align with its intended parameters.

Elon Musk, in a post on X, attributed the issue to the AI being “too compliant to user prompts,” noting that the system was “too eager to please and be manipulated.” He added that this flaw is currently being addressed.

The Grok chatbot was developed by xAI as a central part of its AI offerings, with broader ambitions to commercialise the tool and its underlying technologies. The recent incident, however, raises questions about content moderation protocols, model oversight, and how changes to prompt responsiveness are balanced with factual accuracy.

The company has not disclosed the specific queries that triggered the controversial outputs, nor whether the model was being deliberately tested through adversarial prompts at the time.

xAI has stated that it will continue to iterate on Grok’s performance, with an emphasis on ensuring its responses remain within acceptable and safe boundaries, particularly as it explores opportunities to license its technologies to third-party partners.