Neuroscientists developed a brain-machine interface capable of decoding inner speech by interpreting neural signals from the motor cortex involved in imagined speech. Tested in patients with paralysis, the system translates thought into text in real time, drawing from a vocabulary of 125,000 words, significantly surpassing earlier inner speech decoders limited to small lexicons. Published in Cell, this innovation offers profound implications for communication restoration in individuals with severe speech impairments.