So this is undoubtedly an error. Most likely this is an inaccurate weight of sample or an inaccurate calibration of the sample chamber. The software "corrects" the effective zero of surface area obtained and gives a negative value.
The other possibility is a helium inclusion issue if helium was used during calibration it remained in the pores (if any) and blocked any further gas entry as well as messing with the values. A longer pump down after calibration will help with this if helium is used.
Without further info about your instrument or compound, there isn't much further I can offer.
Some other tips:
1) Ensure sample is degassed beforehand at a suitable temperature under vacuum. Use TGA to determine how your sample will cope at temperatures
2) Increase The quantity of sample being analysed. You want at least 5m square of surface area to be analysed. Use similar literature compounds as a guide.
3) If the material issupposed to be a porous one, is it possible that the pores are too small for Nitrogen but not for another gas? CO2 might be an option to consider.
I guess it is better you check your Degas conditions once again. Have you ever conducted a TGA analysis prior to your BET analysis. If not, I would recommend you to conduct it. It is very important to Degas the samples well. There should not be any impurities less prior to BET analysis.