I'm not a professional and some of this interpretation may be wrong:
In the first report, 83% of people who tested positive for viral RNA (an active infection) had immune T cells that seem able to neutralize the virus, possibly without antibodies. I don't see any information on whether these patients were also tested for antibodies later, but in theory some of them could have cleared the infection without developing any.
Another cohort of healthy blood donors was tested for SARS-CoV-2 antibodies, and among the samples that did not have them, 34% had T cells that also seem able to neutralize the virus. They note that these cells appeared to work in only one of the two ways that were observed with the PCR-test-positive patients' T cells. The idea is that this one method of neutralizing SARS-CoV-2 could be a carry-over from other human coronavirus infections in the past, and any antibodies generated by this method would be ignored by tests specific to SARS-CoV-2.
Also the seronegative healthy donors skewed female and younger compared to the PCR-test-positive patients, but I'm not sure if that is significant.
In the second report, around 5% of people who tested positive for viral RNA (had active infections), and then recovered, still developed no detectable antibodies. It's possible that more than 5% of exposed people are clearing the infection without developing antibodies, because this sample was taken from confirmed cases, and someone with pre-existing immunity may be less likely to develop symptoms.
Of course, keep in mind that both of these are preprints, and may contain methodological errors.