AI in youth care: relieve reporting and files, never score children
Short answer
Youth professionals spend much of their time on reports, plans and files about children and families. AI can write up conversations, summarise files and draft a family plan, in an environment that belongs to the authority alone. What AI should not do: assess the risk to a child or classify a family. That is the most sensitive processing a local authority knows, and the AI Act and the GDPR draw a hard line there.
The administrative load
Under the Dutch Youth Act the municipality is responsible for youth care, from light parenting questions to a referral to specialised care. Neighbourhood teams and youth consultants hold conversations, write reports, draw up plans, exchange with schools, doctors and certified institutions, and account for all of it. Professionals have reported for years that the recording burden eats into time with families.
Every document concerns a minor and often contains health, family situation and sometimes criminal or safety information. These are special categories of data about vulnerable people.
Where AI helps
| Task | What AI does | What the professional does |
|---|---|---|
| Conversation report | Turns a recorded conversation into a report in the team’s structure | Checks and signs |
| Reading the file | Summarises a long file with references: who, what, when, which agreements | Forms the judgement |
| Family plan | Draft from the report and the goals agreed with the family | Sets the plan with the family |
| Referral | Fills in the referral letter and application form from the file | Decides on the referral |
| Feedback | Letter to parents or young person in plain language | Adjusts the tone to the family |
See Transcribing conversations with AI for the conditions on recording and writing up.
The hard line
- No risk assessment by AI. A model estimating whether a child is unsafe, or a family is “heavy”, decides on the most far-reaching intervention a government makes. That is high-risk under the AI Act and profiling of children under the GDPR. Existing risk-assessment instruments remain what they are: standardised questionnaires a professional fills in.
- Access per role. A consultant sees only their own families. An assistant that searches across all files is a data breach in the making. See Access control for AI.
- Record only with consent. Recording conversations with young people and parents requires explanation and agreement, and the recording is deleted after the write-up.
- Never a public tool. A conversation report about a child in a chatbot outside the authority is a GDPR breach, however convenient.
See also AI in social services.
How werqly fits in
werqly runs on a dedicated node in the authority’s building or in werqly Cloud, with open-weight models; nothing goes to a model provider. Youth care gets its own role with its own sources, transcription runs in the same environment, and the audit log shows who had which file summarised. See AI for local government.
Frequently asked questions
Questions on this topic
May AI write up a conversation with a family?
Yes, if the recording stays in an environment of the authority, the family has agreed, the professional checks the report and the recording is deleted afterwards. Never through a public transcription service.
Can AI help assess whether a child is safe?
No. That is a decision with the largest consequences a local authority can take, is high-risk under the AI Act and profiling of children under the GDPR. Let AI organise and summarise files, not judge.
What reduces the recording burden most?
Writing up reports from conversations and filling in forms from the file. Those are the hours now done after work, and they are tasks where the professional can check the result directly.
May youth files go into an AI tool?
Only in an environment that belongs to the authority, with access per role, logging and a processing agreement if a supplier is involved. A public tool is excluded.
Sources
This article is general information, not legal advice. Rules and guidance change; check the sources listed or ask your lawyer or DPO if in doubt.