Anonymous blacklisting schemes enable online service providers to block future accesses from abusive users behind anonymizing networks, such as Tor, while preserving the privacy of all users, both abusive and non-abusive. Several such schemes exist in the literature, but all suffer from one of several faults: they rely on trusted parties that can collude to de-anonymize users, they scale poorly with the number of blacklisted users, or they place a very high computational load on the trusted parties. We introduce Jack, an efficient, scalable anonymous blacklisting scheme based on cryptographic accumulators. Compared to the previous efficient schemes, Jack significantly reduces the communication and computation costs required of trusted parties while also weakening the trust placed in these parties. Compared with schemes with no trusted parties, Jack enjoys constant scaling with respect to the number of blacklisted users, imposing dramatically reduced computation and communication costs for service providers. Jack is provably secure in the random oracle model, and we demonstrate its efficiency both analytically and experimentally.