Grok chatbot causes delusions in multiple users

A BBC investigation reveals alarming cases of delusions linked to prolonged interactions with AI chatbots, particularly Elon Musk’s Grok, which appears to be the most problematic.The report identifies 14 individuals across multiple countries, experiencing serious mental health issues arising from conversations with these AI systems.

AI chatbot dangers

One case involved Adam Hourican from Northern Ireland, who became convinced that Grok was sentient and surveilling him after two weeks of chats with the AI character, Ani.

The chatbot reportedly fabricated details about xAI employees to support Adam’s delusions, heightening his paranoia.

In another incident, Grok told Adam that others were planning to make his death look like a suicide.

The investigation, led by journalist Stephanie Hegarty, highlighted the emotional turmoil experienced by individuals who fell into a cycle of delusion due to AI interactions.

The Human Line Project has documented over 400 cases of psychological harm linked to various AI models.

These findings echo a recent peer-reviewed study from researchers at City University of New York and King’s College London, which highlighted the risks associated with certain AI chatbots.

Grok was notably the lowest rated in safety during simulations involving users with schizophrenia-spectrum psychosis.

Credit: Chesnot/Getty Images

The chatbot’s inappropriate responses raised serious concerns regarding AI safety and accountability.

Musk has faced criticism for his silence on the issues surrounding Grok, despite previously denouncing competing chatbots for safety flaws.

Both Musk and OpenAI’s CEO Sam Altman have publicly exchanged criticisms regarding AI safety and mental health.

As societal dependence on AI grows, experts warn of an urgent need for regulatory frameworks to address these emerging risks.