Person 2 understands person 1 to say: "Vaccines can cause autism, in view of what we know."
Person 1 then seems to claim to have meant something along the lines of: "Vaccines can cause autism, in view of what is conceivable."
Kratzer calls these „in view of“-clauses the conversational background. It determines what words like "can" and "must" mean. One way to explain why we would consider the reasoning of person 1 to be fallacious is by their handling of implicit conversational backgrounds.
If person 1 were to assume the second conversational background, they were certainly right since it is conceivable that vaccines cause autism and scientists cover that up, even when there is absolutely no evidence in favour of this assumption.
But by the mere act of making this statement they suggest that it conveys information of practical relevance. However, for it to be practically relevant, it would have to describe a possibility in view of what we know. Therefore, despite them suggesting that they assume the second conversational background, they act in a way that only makes sense if they assume the first. And when assuming the first conversational background, the statement is plain wrong.