I’m not sure if this is just rumors, or if there’s any hope to succeed. If not a rumor, who knows what exactly might analyzing encrypting data means… Nonetheless, if everybody were using distributed (not even just decentralized) communications and sharing mechanisms…
Homomorphic encryption is still in early development in the academic community. I doubt Facebook is going to boost development majorly. It would ultimately mean breaking the Signal protocol that Signal implemented/audited for them.
Thankfully for now, they can’t learn much except the length of our data, and deduce context of our conversations with metadata.
Holomorphic encryption is not about breaking existing encryption. It’s about building new cryptosystems that allow specific operations that current cryptosystems don’t. The goal is to build a cryptosystem where given a function
f
such thatf(x) = y
you want to have an encryption scheme where someone who doesn’t knowx
but knowsencrypt(x)
can calculateencrypt(y)
without learning anything aboutx
ory
.If someone found a way to do something like that with a current cryptosystem it would be considered broken.