pValue corrections for fitglm?

I have a table with three columns:

1) group ID - categorical ranging from 1-500

2) time point - double from 0-inf+

3) dependent variable - log-normal intensity values

We are testing whether or not, time interacts within each group ID to alter response for that group ID. What we expect to see is no effect. To analyze this data I used:

fit = fitglm(glm_table,['response~group:time']);

Yet, approximately 22% of groups are showing pvalues less than 0.05. I'm thinking this has to do with the number of interactions terms (500) and the high number of degrees of freedom (~36k). Also when I plot, most data points are clustering around the mean for all time point, while there are only a couple of large outliers. So I think these are mostly false positives. How should I go about doing multiple corrections for these pvalues? I've tried FDR, with little effect.