Differential Privacy
A guarantee that a result looks statistically almost identical whether or not any single individual's data was included.
Achieved by adding noise calibrated to how much one person could change the answer. The strength is a parameter, epsilon: smaller means more noise and a stronger guarantee.
Two properties make it the only technique here with a real defence. It holds against an attacker with outside information, so it can't be broken by cross-referencing another dataset — which is exactly how k-anonymity fails. And epsilon is a budget: every query spends some and the guarantee degrades as they accumulate, which is what makes the "ask two queries differing by one person" attack impossible rather than merely awkward.
It costs accuracy, and that's the point rather than a defect. Every other anonymisation technique makes the same trade and doesn't tell you the size of it.