What does the AI Language Decoder do with encrypted communication?

It observes what is legitimately observable — source, destination, protocol and metadata — and reports the payload as unavailable because authorized decryption material is unavailable, with semantic interpretation marked unknown. AI Control does not build or advertise any unauthorized encryption bypass.

Frequently asked questions

Can metadata still be useful?

Yes. Timing, sizes and endpoints can inform risk, but they are not treated as content.