Number of errors in a sample to be not valid

A sample was performed by a third party with a 90% confidence level and with a 10% maximum error rate on a population of 1,000 yielding a sample size of 64

We want to re-test the sample of 64. The question is when we look at each item of the 64 sample, how many errors (Assume the individual sample is either an Column A or Column B sample and we disagree with the individual conclusion) before we can conclude that the sample is not valid

I hope I am clear