A research team from the University of Texas at Austin significantly advanced brain imaging by employing a ChatGPT-like AI model for decoding human thoughts with up to 82% accuracy from functional MRI (fMRI) recordings. The study, published in Nature Neuroscience, demonstrates the potential for further applications in neuroscience, communication, and human-machine interfaces while raising concerns about mental privacy. Researchers gathered 16 hours of brain recordings from three subjects who listened to narrative stories and identified neural stimuli corresponding to specific words. This breakthrough was possible due to custom-trained GPT Large Language Models (LLMs) that generated intelligible word sequences from perceived speech, imagined speech, and even silent movies. The AI model accurately deciphered both the meaning of stimuli and specific words the subjects thought. The study's success raises concerns for mental privacy. Researchers found that subjects could resist decoding efforts by employing techniques such as counting or listing animals, which significantly decreased accuracy. The scientists acknowledge the risks of brain decoding technology and suggest raising awareness and enacting policies to protect mental privacy.