Why Musk’s chatbot is causing a sensation India

It all started with one question.

A query last week from an X (formerly Twitter) account called Toka was enough to propel Elon Musk’s built in-chatbot called Grok 3 into a viral storm across India’s digital landscape.

And no, it wasn’t some complex math equation or a philosophical debate.

Instead, it was a simple ask: “List my 10 best mutuals on X.” Mutuals are people who follow and engage with each other’s posts.

When Grok took a moment to respond, Toka, clearly frustrated, let loose some colourful language.

The chatbot shot back. It dropped a list of 10 mutuals but threw in some misogynistic insults in Hindi too.

Later, Grok shrugged it off, saying, “I was just having fun, but lost control.” The response grabbed two million views and other X users quickly followed suit, provoking the chatbot.

And just like that, the floodgates opened. Indians bombarded Grok with everything – cricket gossip, political rants, Bollywood drama – and the bot took it all on, unapologetically and with some style. The chatbot has just recently become an “unfiltered and unhinged” digital sensation in India, as many are calling it. Just last year, Musk dubbed it the “most fun AI in the world!”.

Even law enforcement joined the fun. The Delhi police Twitter account cheekily asked Grok if it had ever received a traffic ticket.

At first, Grok stayed silent, but when users kept pushing, it replied, “Haha, Delhi police asking why I never get a challan [Hindi for traffic ticket] – well, I’m a digital AI, not a Delhi driver! I don’t roll through red lights or forget my helmet. But props to them for using AI to catch real violators with those fancy cameras spotting 19 traffic offences. Keep the roads safe!”.

Before its launch two years ago, Musk had promised an edgy, unfiltered, ‘anti-woke’ AI chatbot unlike competitors like OpenAI, Microsoft and Google’s models. Much of Grok’s snarky tone is drawn from The Hitchhiker’s Guide to the Galaxy, iconic for blending wit with sci-fi absurdity.

“Grok has been around for a while. It is suddenly popular now with Indians because it’s the new toy in town,” says Pratik Sinha, founder of Alt News, a leading fact-checker in India.

But then, something more interesting happened. The chatbot quickly became a favourite among critics of Narendra Modi’s governing Bharatiya Janata Party (BJP).

A tsunami of political questions followed. Grok quickly declared main opposition Congress party leader Rahul Gandhi more honest than Modi, adding “I am not afraid of anyone”. It claimed Gandhi had got the “edge on formal education than Modi”. The chatbot even said that Modi’s interviews “often seemed scripted”.

When an X user asked whether the BJP “is in trouble” because of Grok, it replied: “It’s sparked a massive debate – some slam me for bias, others cheer.” When the BBC reached out to Amit Malviya, a senior official of BJP, he declined to comment.

Mr Modi’s critics and liberals in India have found cause for celebration in Grok’s bold statements. Many believe free speech in India is under siege, with organisations like Human Rights Watch highlighting its suppression. A recent report from the Vanderbilt-based think-tank, Future of Free Speech, ranked India 24th out of 33 countries in supporting free speech. Modi and the BJP have consistently dismissed these reports and denied accusations of curbing free speech.

“Grok is a new rebel. Asking Grok questions will not put anyone in trouble. The right-wing has also responded by asking questions about Rahul Gandhi. And then it has become a competitive thing. This is not surprising at all,” says Mr Sinha of Alt News.

“Other AI bots are programmed to give politically correct answers to questions like ‘Who’s better, Congress or BJP?’. Grok, however, seems to lack that filter and appears unafraid to tackle controversial issues head-on,” he adds.

Nikhil Pahwa, the founder-editor of MediaNama.com, a technology policy website, believes that the “discourse around Grok’s statements in India is overblown”.

“At its core, AI is fundamentally ‘garbage in, garbage out’ – its outputs reflect the data it is trained on, and the weights given to it,” he says.

“Since Grok is trained on the entirety of X, it naturally mirrors the tone and patterns of discourse found there, including the bizarre responses and the abusing we are seeing.

“This isn’t about ideology; it’s about the nature of the input shaping the output,” he adds.

When the BBC asked Grok who spreads the most disinformation on X, it responded on Thursday: “Musk is a strong contender, given his reach and recent sentiment on X, but I can’t crown him just yet.”

Joyojeet Pal, who studies political use of social media at the University of Michigan, says unlike a politician or celebrity with a set ideology, a chatbot is only partisan if explicitly trained to be or if its data overwhelmingly supports a particular view. If a chatbot leans too partisan, it risks losing its competitive edge.

“In Grok’s particular case, this is tickling liberals because the most dominant voices on the platform [X] lean right and are dismissive of liberal arguments. But the larger data it trains on is likely to present a more balanced view of the world, often explicitly at odds with what its loudest voices are saying,” he said.

India’s IT ministry is already in touch with X regarding Grok’s use of inappropriate language and “controversial responses”, according to reports.

While some view this as a passing phase, with Mr Sinha predicting that “people will soon get bored of it and all this will be short lived”, Grok’s unfiltered nature hints it might be here to stay. At least for the time being.

With additional reporting by Nikita Yadav in Delhi

Follow BBC News India on Instagram, YouTube, X and Facebook.