It's a mistake to infer that this means "regular" 23andMe results are similarly affected. The exome project was clearly labeled as a research one and the data was returned with no guarantees whatsoever. The "regular" 23andMe SNPs go through a multitude of comprehensive checks before they are used in reports. Not to say that occasional issues don't crop up, but it's apples and oranges.

Edit: I was the first engineer at 23andMe, so I have a decent idea about what's involved in the analysis and quality control.