I can only think of 2 things off the top of my head
(1) somehow you input the same group of subjects twice, so you are getting 'A-A' which is always 0. This is easily checked by looking at the individual set results from 3dGroupInCorr
(2) somewhere in the processing (before or during 3dSetupGroupInCorr), all the datasets got turned to zero. If you use the '-sendall' option to 3dGroupInCorr, then all the individual subject results are sent to AFNI as well as the group results. Are those all zero as well?
(3) Evil spirits have taken over AFNI, and exorcism is needed.