An inner-speech decoder reveals some mental privacy issues (arstechnica.com)

Researchers at Stanford University have developed a brain-computer interface (BCI) capable of decoding inner speech—the silent, internal monologues we experience when reading or thinking—without requiring physical attempts to speak. This marks a significant advance over prior BCIs that relied on decoding signals tied to muscle movements, which are exhausting or impossible for patients with severe paralysis, such as those with ALS or tetraplegia. By focusing on neural activity associated with silent speech, the team aims to make communication more accessible for these individuals. The innovation comes with an important caveat: inner speech often contains deeply private thoughts, raising concerns about mental privacy. To address this, the researchers introduced a novel "mental privacy" safeguard designed to prevent the unintentional decoding of unintended or confidential internal content. Technically, their system uses microelectrode arrays implanted in the motor cortex to capture brain signals related to silent speech, sidestepping the need for muscle movement signals typically leveraged in speech BCIs. They trained AI algorithms on data collected from four nearly paralyzed participants performing tasks like listening to words and silent reading, showing initial success in translating neural patterns into meaningful word interpretations. This work opens a promising avenue for enabling communication in patients unable to speak but also highlights the ethical challenge of protecting users’ mental privacy. It underscores the delicate balance in neurotechnology between unlocking new capabilities and safeguarding intimate cognitive experiences.