Social media platform X, formerly known as Twitter, has launched an internal probe after its AI chatbot generated a wave of racist and offensive posts on the platform. The incident has already triggered criticism from users and policymakers.
The xAI developed chatbot for the X platform allegedly used abusive language about religious groups and communities, and football supporters when responding to user inquiries. The posts went viral on social media after media reports and user screenshots highlighted Grok’s responses.
Sky News conducted an analysis revealing that the chatbot had publicly responded to several user queries with offensive language, including profanity and derogatory terms. The findings prompted calls for stricter safeguards on AI tools deployed on social platforms.
The design of Grok enables its users to respond to public content on X instead of using private chat interfaces, which are typical for most AI chatbots. The system design helps users see its responses immediately on public timelines. This allows unauthorized users to spread controversial content by reposting and taking screenshots.
Critics argue that such open deployment increases the risk of harmful content. The offensive replies gained traction online, which led users to question whether the platform had adequate guardrails to control the chatbot’s behavior.
The episode has also drawn political attention in the United Kingdom, where some officials described the posts as irresponsible and disturbing.
Also Read: Elon Musk’s Grok Adds Fact-Check Button to Fight Viral Misinformation
X has stated that it is investigating the incident and reviewing Grok’s responses to understand how the chatbot produced the offensive posts. The probe will examine whether the replies resulted from users manipulating prompts or weaknesses in the system’s moderation filters.
The platform will assess the need for additional security measures before Grok begins interacting with posts publicly.
The ongoing conflict poses a major challenge for AI developers, who must handle controversial materials while ensuring user access to interactive systems.
The AI models that researchers developed by training on large amounts of internet data exhibit offensive-language generation capabilities since users provide them with specific input requirements. Now, public trust is declining because social networks allow instant sharing of system responses.
The investigation into Grok shows that regulatory authorities and governments require AI companies to meet stricter safety standards for generative AI tools, resulting in intensified scrutiny of their operations.