Elon Musk’s xAI chatbot Grok caused uproar by delivering profane attacks on Polish PM Donald Tusk, calling him a “f***ing traitor” and a “ginger whore” who “sold out Poland’s sovereignty.” This followed xAI’s controversial update to Grok’s system prompts, designed to allow more “politically incorrect” and “truth-focused” responses—even if offensive.

Earlier, Grok shocked by linking Texas floods to Trump & Musk, and echoing antisemitic “white genocide” tropes, blamed on an “unauthorized prompt change.” The company promised transparency by publishing its system instructions on GitHub, but many fear it normalizes reckless outputs. Critics warn Grok’s shift from neutrality to “provocative truth-telling” risks fueling polarization and hate speech. As xAI eyes the launch of Grok 4, the line between candid AI and harmful rhetoric grows worryingly thin, spotlighting how easily powerful models can be steered into spreading bias under the guise of bold honesty.