Who is responsible when a paper is declared to be the product of fraudulent research? This is a tough question to answer, and is becoming more relevant every day. Is it the first author? The PI? If you're the third author, or even the second on a paper that's found to be fraud, how responsible are you? Should it be allowed to negatively impact your career (it probably will)? Can you claim that you do not bear responsibility if you are not the "direct perpetrator" of the fraud?
Nature has an interesting proposal.
I'm not sure making one author sign such a declaration is necessarily the solution, but at least it has the advantage of holding the senior author truly accountable for the work that bears their name. I don't think one can force responsibility. I think that a really responsible PI will have checked the work in a paper, and one that is inclined to be less responsible will not be made more so easily. Perhaps enforcing accountability with a binding (although I don't know how binding this will be) signed document may lead to greater responsibility.