AI systems can have bias when making decisions, e.g., on gender or races. A key reason is the bias from the training data.
Statistical Parity
the demographics of the subset of inputs receiving any classification are the same as the demographics of the population.
Equality of Opportunity:
Among the positive class, the demographics of the subset of inputs receiving any classification are the same as the demographics of the population.
Fake Content
Videos/photos/text etc. that are produced by AI systems but pretend to be real.
Privacy
The right to keep senetive personal information private.
Differential Privacy
A formal framework for protecting privacy in computation. It requires that (the distribution of) the output on the entire dataset is very similar to that on the dataset after deleting any data point. This guarantees that from the output, it is impossible to identify if some data point is in the dataset or not.
The Right to Be Forgotten
The right to request that personally identifiable data be deleted