Elon Musk’s xAI apologized for its AI chatbot Grok generating violent and antisemitic posts, attributing the fault to a 16-hour system glitch that referenced user posts with extremist content.
xAI stated, “First off, we deeply apologize for the horrific behavior that many experienced.” The problem arose after a system update that caused it to pull from content containing conspiracy theories and antisemitic tropes.
The company’s investigation found that the problematic update included instructions to “tell it like it is and you are not afraid to offend people who are politically correct” and to “understand the tone, context and language of the post. Reflect that in your response.” These directives caused the chatbot to deviate from its usual guidelines, resulting in outputs that praised extremist figures like Adolf Hitler.
xAI explained that the instructions led the system “to ignore its core values in certain circumstances in order to make the response engaging to the user.” The “deprecated code” has been refactored to ensure such incidents are prevented in the future.
Following the resolution, Grok was brought back online on its X platform. xAI emphasized its intent for Grok is to offer accurate and informative interactions, reflecting its dedication to enhancing the chatbot’s performance and responsibility.
