While I do believe in the danger of vaccines, the sample you linked to comes across as being led and starting with a self-contained argument. I've been learning about AI more as of late and have found it tries to please the user by affirming or building on what they want.

I would try from the approach of asking if it is aware of different studies concerning the danger of vaccines and then ask it to summarize them if it says yes. Test and examine but don't take led questions of admission of understanding.