Therapy is built on a relationship between people.

It's Complicated exists to create the conditions for therapeutic encounters between people. We do not provide AI therapists or chatbots that simulate a therapeutic relationship.

When someone comes to It's Complicated looking for therapy, our goal is to help them find the right human therapist and provide both client and therapist with safe tools to support the process.

There are many things around therapy that technology can make easier: finding a therapist, administration, scheduling, invoicing, organising a practice, and potentially supporting therapists and clients in other practical ways. We distinguish these uses from the therapeutic relationship itself.

AI should be a tool people knowingly choose to use, not an invisible participant in the therapy room.

Therapy involves some of the most sensitive information a person can share. That is why clinical data on It's Complicated is designed to remain private, secured by encryption by default and stored in GDPR-compliant European data centres. Beyond this, we continually strengthen the platform with additional security features for therapists to protect their records and practice.

By default, we will also not make this data available to AI systems. If we introduce features that allow AI to be used with clinical information, this will require an explicit choice and authorisation. Where such use concerns information about a client, we believe clients should also be appropriately informed about how their information is being processed.

AI involving clinical data will always be opt-in, never opt-out.

Where an AI-supported feature is explicitly used with clinical information, we will only work with systems and technical arrangements designed to prevent that information from being retained or used to train AI models.

What a client shares in therapy should not quietly become material for training artificial intelligence. We believe confidentiality should extend to the technologies surrounding the therapy room.

People should know when they are interacting with AI or when AI is being used to process information concerning them.

We do not believe AI should operate invisibly behind experiences that people reasonably assume are human. Where we use AI, we will make that use understandable and transparent — whether the person using our platform is seeking therapy or providing it.

We will also clearly identify AI interactions and AI-generated content where required under the EU AI Act.

AI can generate, organise and suggest. It cannot assume the responsibility of a therapist. We design AI-supported tools as aids to professional judgment rather than substitutes for it, and important clinical decisions should remain with qualified human practitioners.

Clients should also be able to trust that the care they receive is ultimately shaped by the therapist in front of them, not delegated to an algorithm.

AI will change, and so will the questions surrounding it. A useful policy therefore cannot simply be a list of technologies we allow or prohibit today. Our guiding question is: Does this technology support the human relationship at the heart of therapy, or does it diminish it?

That means considering both sides of the relationship: the therapist's ability to practise well and the client's ability to enter therapy with privacy, trust and confidence that another person is genuinely there with them. When the answer is unclear, we believe caution is warranted.