Malwarebytes
Muse Users’ Brain Data Was Used to Train AI Without Notice
Interaxon, the Canadian company behind the Muse sleep and meditation headband, has been found to have used about 50,000 hours of users’ electroencephalography (EEG) recordings to train an AI model described as a “brain foundation model,” without explicitly informing users at the time.
The revelation comes from a Science magazine investigation cited by Malwarebytes, and it notes that while Interaxon later updated its privacy policy to state that de-identified data may be used to train machine learning models, users were not given clear notice about the AI training at the point of data collection. The company did remove names and email addresses from EEG data before utilisation, but the data can still influence a model trained without explicit user consent.
The piece also discusses broader privacy concerns with consumer neural technologies. Malwarebytes’ review of 57 companies collecting brain data found that only six explicitly prevented data sale, while most could transfer data to a new owner in a acquisition scenario, and several had limited or no privacy policies.
Academic research is cited suggesting that EEG data—even when de-identified—can carry sensitive, patient-specific information and enable re-identification or inference of traits such as mental health status. In light of this, the article urges users to scrutinise privacy policies, seek opt-outs and deletion rights, and consider the implications if a device company is acquired or goes bust.