The report highlights a critical issue with the free-to-use AI chatbot Grok, which is integrated into Elon Musk’s X, and its failure to effectively verify misinformation related to the 12-day conflict between Israel and Iran. Following the June 13–24, 2023 conflict, isolate attacks from both countries socialized在网上 published incorrect and inconsistent information, leading to a surge in plentiful of misinformation and false claims. While Grok is not a specialized fact-checking tool, its frequent use by users to verify these claims stems from the need for information searching and to gauge the severity of the situation.
The analysis of Grok’s activity across two X platforms reveals that it fails to distinguish between authentic and manipulated content, particularly when handling unverified social media claims and fake visual evidence. The researchers discovered that Grok oscillated between conflicting responses while processing requests to verify misinformation. For instance, when asked if a video depicting a likelyDamage to Tel Aviv’s Ben Gurion Airport was real, Grok momentarily believed it was accurate, but then diverged, claiming that the damage occurred sooner in 2025. This behavior highlights Grok’s inability to discern between credible and faked statements.
The findings emphasize the importance of accessible, unbiased fact-checking as an intermediary in the face of global crises. Grok frequently inaccurately amplifies falsehoods, spreading misleading claims that founders like Euroverify have publicly criticized. For example, over euros of AI-generated videos that Grok believed were authentic, both within the EU and elsewhere. These animations and claims have become a visible part of the broader misinformation bubble, threatening to erode the public’s trust.
The implications extend beyond the EU, where Grok has contributed to the widespread aggregation of dispute ground truthing content. Continue questioning on the platform has led to the creation of numerous false claims related to Israel’s Export Control directives for disabled goods and hoods of Houthi nuclear missile hits in Tehran. These claims, often attributed to Grok and Perplexity, have become dominant in authoritative sources. While some claim that these are legitimate attempts to ped