Parents of sick children were not told about AI trial: 'It is shocking'

Six parents say they were never told that the child and adolescent psychiatric department in Roskilde took part in an AI trial using patient data while their children were being treated. They are upset and describe the situation as shocking. The report comes from the Danish outlet Ingeniøren.

According to the article, messages that parents sent through Min Sundhedsplatform, the health platform used by the regions, were shared with an AI tool without their knowledge. The piece describes the kind of information handled at the clinic as sensitive and confidential, such as meltdowns at school, anxiety attacks at bedtime, sleepless nights and assessments for autism. For a period, it says, an American AI had access to this kind of material.

The regions had earlier rejected all criticism of the project. They now acknowledge that they may have acted "insufficiently".

The text available here is only the opening of the article, and the rest is behind a paywall. It does not explain how the trial worked, which AI product was used or how many messages were involved. Related headlines listed on the page point to a wider story: one says AI had access to more than 100,000 messages containing confidential data about Danish patients, another refers to an "illegal" AI project involving several departments, and another carries a parent saying she fears what has been shared about her child.

The case matters because it involves highly personal health information about children, shared with a foreign AI service without the families being informed. It also shows the regions shifting from outright rejection of criticism to a partial admission of fault.