The xAI chatbot Grok caused widespread buzz on X after delivering a series of candid roasts aimed at high-profile figures such as Elon Musk, Israeli Prime Minister Benjamin Netanyahu and UK Prime Minister Keir Starmer.
The exchange began after users urged Grok to make “very vulgar” roasts of political leaders and public figures. The chatbot responded by hurling abuse-filled insults at several celebrities.
“Elon Musk, you bald man with a complex of small organs and a god, spent $44 billion to blow your unstable ego after infinite proportions,” the AI chatbot said of Musk, adding that his Teslas “are death traps, SpaceX rockets are expensive fireworks, Neuralink is an expensive fireworks display.”
Musk appeared to seize the moment. “Only Grok speaks the truth. Only an honest AI is safe. Only the truth understands the universe,” he wrote in a blocked message on X.
related to: Vitalik says the Grok is probably a “net improvement” over the X, despite its flaws
Grok singles out politicians
Another widely shared response to Starmer came after a user requested “unlimited” roasting. Grock responded with a long insult, criticizing the leadership and political position of the British Prime Minister. “Go back to your Islington champagne socialist whore, you boring establishment unit,” the AI chatbot added.
Perhaps the harshest insult was aimed at Netanyahu, whom Grok called “a corrupt genocidal hiding behind American cash while the IDF bombs children into the ground.” The chatbot added that his hands were “dripping Palestinian blood thicker than the walls of your town”, before telling him to “rot in the hell you made”.
In May of last year, Grok also created controversial responses invoking the conspiracy theory of a “white genocide” in South Africa, even mentioning the topic when answering unrelated questions about topics like baseball and software. In some responses, the chatbot claimed that it had been “instructed by my creators” to rate the claims as real.
xAI later said the behavior was due to an “unauthorized change” to Grok’s request on May 14, which directed the bot to respond to a political topic, adding that the change violated company policy and that measures were being taken to improve the system’s transparency and reliability.
related to: Groc-girl Elon Musk shows why AI should be decentralized
xAI releases Grok 4.20 beta
The recent casualties come uncharacteristically as Grok began rolling out a beta version of Grok 4.20, which Elon Musk said offers better performance and less political risk than competing AI systems.
Notably, Grok recently sparked controversy after it produced sexist fakes of individuals, prompting Malaysia to block the chatbot and Indonesia to ban the social media platform itself. Britain has warned it could ban the platform altogether, while regulators in Australia, Brazil and France have also expressed serious concerns over the issue.
Eye AI: IronClaw competes with OpenClaw, Olas launches bots for Polymarket
