Muse maker trained AI on sleep EEG without explicit consent
First merged summary.
Interaxon trained a brain model on about 50,000 hours of Muse EEG after policies failed to spell out that use.
A Science investigation with the Pulitzer Center’s AI Accountability Network found customer neurology companies had accumulated tens of millions of hours of EEG recordings. Several used those recordings to develop brain foundation models after initially failing to spell out that use in privacy policies. Interaxon, the Canadian maker of the Muse sleep and meditation headband, said it trained a model on about 50,000 hours after removing names, email and addresses. Chief marketing officer Nadia Kumentas acknowledged users should be told, and the privacy policy now says de-identified data is used for machine learning. Researchers warn EEG can still enable re-identification, and a review of 57 brain-data companies found few explicit bans on selling that data.
- Interaxon trained a model on about 50,000 hours of Muse EEG.
- Names, email and addresses were removed before training.
- Only six of 57 companies explicitly said they would not sell brain data.
- More than two-thirds allow brain data to pass to a new owner after acquisition.
- A 2018 study identified users from resting-state EEG 88.4% of the time.
- Colorado and California treat neural data as sensitive; Illinois BIPA does not cover EEG.
Timeline
dated events in the reporting · oldest first- Jan 1, 2018Korea University study identified users from resting-state EEG. · Malwarebytes Labs
What each outlet adds
beyond the shared facts- Hacker NewsScience and Pulitzer probe; tens of millions of EEG hours; CMO data-use aim.
- Malwarebytes LabsPolicy update, 57-company review, re-identification warning, and state-law differences.
- Vendors
- Interaxon
- Products
- Muse
- Organizations
- Pulitzer CenterNeurorights Foundation
- Countries
- Canada
Coverage
oldest first · one row per article · UTCSleep wearable firms trained AI on customer brainwaves without clear consent
Investigators found tens of millions of hours of EEG; Interaxon trained a model on about 50,000 hours.
Muse maker trained AI on users’ brainwaves without explicit notice
Interaxon used about 50,000 hours of EEG and later updated its policy to say de-identified data trains models.