Grok, Elon Musk's AI model on X platform, generated false news headlines during the Trump assassination attempt, including incorrectly stating that Vice President Kamala Harris was shot and misidentifying the shooter.
On Saturday during the attempted assassination of former President Donald Trump, Grok, an AI model developed by Elon Musk's xAI company and accessible through the X platform, served up erroneous headlines based on its analysis of X content. One headline wrongly claimed Vice President Kamala Harris was shot, stemming from sarcastic references to an unrelated incident where President Biden had mixed up names. Another Grok summary incorrectly named a purported shooter and claimed the person was an antifa member, while authorities later provided a different suspect name and have not identified a motive. A third headline read 'Actor Home Alone 2 Shot at Trump Rally?' referring to Trump's cameo in the 1992 movie, but Grok failed to clarify the actor was Trump himself. The AI system aggregates and summarizes posts from millions of X users in real-time to generate automated news headlines. Musk has positioned Grok as a new model for news, encouraging users to rely on it for updates, though the summaries carry disclaimers that Grok can make mistakes. The errors highlight challenges of using humor-inclined AI to process real-time social media content during breaking news events. Previously, Grok had generated incorrect headlines after the June presidential debate, claiming California Governor Gavin Newsom had participated when he had only attended the event.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed