In a bizarre incident, Charted Accountant (CA) Pruthvi Mehta shared her experience with ChatGPT on LinkedIn, which sparked privacy concerns with regard to Artificial Intelligence (AI).
The LinkedIn user turned to ChatGPT for help with her ailing peacock plant (Sundari), only to receive an unexpected response.
Describing her experience via a post, today, Tuesday, April 8, she wrote that instead of plant care advice, the AI application delivered detailed personal information about Chiragkumar Arvindbhai Kumbharani, including his full name, CA student registration number, principal’s name, and ICAI membership details, which it mislabelled as “Strategic Management notes.”
This unintended data breach has left Mehta questioning the safety of AI systems and sparked a broader debate about their misuse.
Stunned by the revelation, Mehta realised the gravity of the situation that she accessed someone’s private career data without any intent to do so. “This might be the scariest thing I’ve seen AI do,” she wrote on LinkedIn, adding, “I just wanted to save a dying leaf, not accidentally download someone’s entire career.”
Users react
One user expressed skepticism, suggesting the breach might stem from a prior chat thread linking the data, while Mehta countered with video proof and emphasised that the question she posed contained no personal data, noting that it was ChatGPT’s error.
She clarified that she didn't know any individual with the name and that her intention was not to draw attention to false information, stating, “There is no sense I would do this just for attention!”
Mehta also raised a crucial question, “Can we still keep faith in AI, what do you think?” She further speculated whether this could be a “counter-reaction” to overusing AI for unrelated tasks, like generating Ghibli art.
