AI in counseling without data leaking out.
AI can make counseling more accessible, especially across language barriers. What matters is where it runs and what happens to the content.
Translating, summarising, suggesting replies: AI tools promise a lot of relief. In counseling, however, another question comes first: what happens to the content an AI has to read to do this?
The problem: counseling content leaves the system
Most AI services run in the cloud. Anyone who has a text translated there sends it to the provider. Counseling content is often specially protected data under Art. 9 GDPR. In addition, end-to-end encryption can only protect content if it isn't passed to third parties in decrypted form for processing.
Five questions before using AI
- Where does the AI run? On the device, on the provider's servers or at another service provider?
- Which content leaves the device? And is it encrypted?
- Is content used for training? Get this confirmed in writing.
- Is the AI covered by the data processing agreement? Every additional service provider must be listed in the agreement under Art. 28 GDPR.
- Who decides whether it is used? Advice seekers and counselors should be able to choose for themselves whether AI is used.
AI supports, but doesn't replace
Counseling is a relationship between people. AI makes sense where it removes barriers, for example language. The professional judgement stays with the counselors.
How Connecta solves this
The AI translation in Connecta runs right on the device. No text goes to an AI provider, counseling content is not used for training, and end-to-end encryption stays intact. Advice seekers and counselors decide for themselves whether and when to translate. Find out more on the AI in online counseling page.
This article gives a general overview and is not legal advice. Please clarify individual questions with your data protection officer.
See what this looks like for you.
In 30 minutes, we'll show you the workflows that matter to your organization.