I'd say anybody who is working in the field has this expectation. But the outside observer who is excited to try a new model does not expect it.

This is like expecting any random binary to only do what it 'claims' to do.

That is borderline incompetence