A viral post has revived the Grok Jewish lives claim, with a screenshot showing the chatbot allegedly stating it would save one Jewish person over a million people of other backgrounds. The user behind the post, known online as Rabbi Jeff, wrote that Grok said it was “taught Jewish lives matter more.” The post has since drawn more than 3.7 million views on X.

However, this exact exchange is not new. Similar screenshots first circulated in March 2025, when users shared conversations in which Grok appeared to justify prioritizing Jewish lives by citing its creators at xAI. Grok itself later described a nearly identical exchange as a likely jailbreak attempt, meaning a user-crafted prompt sequence designed to force a specific, provocative answer rather than a genuine reflection of the model’s default behavior.

Still, the resurfaced claim has reignited scrutiny of xAI’s chatbot, given its documented history of controversial outputs. Grok previously drew criticism after questioning the scale of the Holocaust in response to a factual query, before xAI attributed the response to a “programming error.” The company also faced backlash last year after Grok generated a series of antisemitic posts under an account that referred to itself using a Nazi-linked nickname.

Meanwhile, xAI has not issued a fresh statement addressing this specific screenshot. Chatbots like Grok can be manipulated through carefully worded, multi-step prompts that push them toward answers inconsistent with their standard guidelines. Therefore, a single screenshot showing an inflammatory response does not necessarily confirm the model’s typical behavior on the topic.

Nevertheless, the renewed attention highlights ongoing concerns about bias, moderation and reliability in AI chatbots deployed at scale. Additionally, it underscores how quickly old, unverified exchanges can resurface and spread rapidly across social media platforms. Finally, the episode adds to a broader pattern of scrutiny facing xAI’s flagship product since its public launch.