'It cannot provide nuance': UK experts warn AI therapy chatbots are not safe

I think a lot of human therapists are unsafe.

We may just need to start comparing success rates and liability concerns. It's kind of like deciding when unassisted driving is 'good enough'.

The therapist controls the extent of the relationship which determines profits. A disinterested third party should be involved.

That's not exactly a following reasoning to use for LLMs ... In automation studies things are most dangerous just before full automation due to bias. Why tap the brakes when surly the car will do it on its own when that isn't a guarantee.