Even when implementers adopt a particular SRFI, I would be careful to deduce any kind of endorsements. Some SRFIs are so easily adoptable that implementers include them just for the sake of shipping more SRFIs. Others are included to make the life of users of the implementation easier. And finally, some SRFIs that would be endorsed by an implementer are not shipped because the implementation lacks some features to support this SRFI.
As the final version of a SRFI is solely the responsibility of its author or authors and is not the result of a formal democratic process, they shouldn't be viewed as being endorsed by the community.
Of course - and here I agree with Vladimir - there is always the danger of misunderstanding. But this has to be communicated and solved somewhere else and shouldn't prevent anyone from submitting and/or finalizing a SRFI (unless the author has come to the conclusion - maybe due to discussions with the community - that their idea is not as good as they initially thought).