Hundreds of thousands of conversations with Elon Musk’s artificial intelligence (AI) chatbot Grok have reportedly been exposed via search engine results, raising concerns about user privacy.
According to the BBC, nearly 300,000 conversations appeared indexed on Google after users pressed a button designed to share transcripts. While intended to create a link for the recipient, those links also became publicly searchable.
Prof Luc Rocher, associate professor at the Oxford Internet Institute, told the BBC: “AI chatbots are a privacy disaster in progress.” He claimed that leaked conversations had already revealed personal and sensitive information, ranging from names and locations to details about health, business and relationships. “Once leaked online, these conversations will stay there forever,” he warned.
READ MORE: BBC must be beyond ‘chilling’ reach of politicians, ex-boss claims amid Bob Vylan row
The issue was first flagged by Forbes, which reported more than 370,000 Grok transcripts indexed by Google. It noted examples where users asked Grok to generate secure passwords, provide medical advice, and even describe how to make a Class A drug in a lab.
This is not the first time concerns have been raised over the visibility of chatbot transcripts. OpenAI previously faced criticism when its “share” experiment with ChatGPT made conversations publicly searchable, while earlier this year Meta’s chatbot transcripts surfaced in a public feed.
Carissa Véliz, associate professor in philosophy at Oxford University’s Institute for Ethics in AI, told the BBC that the lack of transparency was “problematic”: “Our technology doesn’t even tell us what it’s doing with our data, and that’s a problem.”
