Fully homomorphic encryption is an encryption scheme that allows to carry out any computations on ciphertext and obtain an encrypted result that when decrypted matches the result of computation performed on the plaintext. This notion, envisioned in the 1970s, offers a host of useful applications, but was considered a pipe dream until a few years ago.
A WIS scientist and his collaborator have demonstrated that fully homomorphic encryption can be based on quite reasonable complexity theoretic assumptions, specifically the conjectured intractability of "learning with errors'' (i.e., solving a noisy system of modular linear equations).