K-anonymity is where you determine that you need at least K individuals in a group with a certain number of shared personal information such that you can't identify individuals within that group. For example, people >2m tall with January birthdays who's favorite color is red. If there was only one person in that group then you could identify them, so you shouldn't use that grouping. But if there are 1000 people in that group (in your dataset), then you can't ID them, so it's okay.