Online targeted advertising has been subject to many complaints for allowing advertisers to discriminate users belonging to sensitive groups, that led for instance Facebook
to disallow the use of sensitive features such as ethnic affinity in certain types of ads. In this talk I argue that such measures are not sufficient and that the problem of discrimination in targeted advertising is much more pernicious. I show how a malicious advertiser can create discriminatory ads without using sensitive features. I argue that discrimination measures should be based on the
targeted population and not on the attributes used for targeting and I propose ways to mitigate discriminatory targeting.