Boundary of perfectly secure classical-client blind quantum computation

Characterize the boundary between Clifford circuits and universal quantum computation at which a classical client becomes insufficient for perfectly secure, circuit-hiding blind quantum computation, including whether circuit classes beyond the identified non-Clifford class can be delegated with perfect security by a classical client.

Background

Blind PBC demonstrates that a fully classical client can delegate, with perfect security, certain circuit classes that include circuits with arbitrarily many non-Clifford gates and are believed to exhibit quantum advantage. This extends the classical-client regime beyond Clifford circuits, for which the QCED protocol defaults to a classical client.

The paper does not determine the full boundary between Cliffordness and universality: it remains unresolved which circuit classes can support perfectly secure, input-hiding blind quantum computation with a classical client. The identified class provides evidence that efficient classical simulability is not necessary, but does not settle the limits of classical-client BQC.