Background:
Let's discussed based on published symmetric white-box crypto only, such as Chow's white-box AES.
I only know basic concepts/objectives of homomorphic encryption, such as PHE and FHE. Bring ciphertext into calculation instead of decryption.
Without proper implementation design for real applications, white-box crypto may reveal its own vulnerabilities, such as code lifting (call whole WB function instead of unboxing WB).
Question:
I am thinking (just brainstorm first) how to bring the homomorphic feature into white-box crypto design. For examples:
A1. Use encrypted plaintext P_IN as white-box crypto encryption input to generate encrypted ciphertext C_OUT.
A2. Use encrypted ciphertext C_IN as white-box crypto decryption input to generated encrypted plaintext P_OUT
A3. Treat white-box as an API module. Use encrypted input/output for further calculation steps.
Or
B. Apply homomorphic features on the intermediate values of white-box crypto, in order to prevent cryptoanalysis.
Remark: PHE or FHE supports limited operations (e.g., addition, multiplications), which could be the potential challenges for white-box crypto.