# Encryption of Magic Backups

**URL:** <https://community.foundation.xyz/t/encryption-of-magic-backups/1236>\
**Category:** Passport Prime\
**Created:** [September 6, 2026, 8:47pm UTC](https://community.foundation.xyz/t/encryption-of-magic-backups/1236 "2026-09-06T20:47:10Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![mailer45](https://community.foundation.xyz/letter_avatar_proxy/v4/letter/m/3bc359/32.png) [@mailer45](https://community.foundation.xyz/u/mailer45)\
**Post date:** [September 6, 2026, 8:47pm UTC](https://community.foundation.xyz/t/encryption-of-magic-backups/1236/1 "2026-09-06T20:47:10Z")

</div>

I was just wondering how safe you consider the magic backups in the long term, as they will in many cases contain a lot of metadata (usernames, passwords, notes, multisig setup) that would most likely link identity to the related bitcoin addresses, etc. I know that the backups are e2e but it’s based on the seed phrase and thus not quantum resistant encryption. Imo anything that is stored online will sooner or later be “stolen” by criminals/governments etc., and once the tech is there in the future they will decrypt it. How well secured are the foundation servers, and is this a valid concern I have or would you say this is too paranoid?

---

<div class="post-metadata">

**Author:** ![qna](https://community.foundation.xyz/user_avatar/community.foundation.xyz/qna/32/284_2.png) [@qna](https://community.foundation.xyz/u/qna)\
**Post date:** [September 7, 2026, 5:03am UTC](https://community.foundation.xyz/t/encryption-of-magic-backups/1236/2 "2026-09-07T05:03:37Z")

</div>

Honest carve out - I’m not a cryptography engineer so I used the help of a frontier model trained on our code base to help me answer this for you.

Magic Backups contains two distinct elements. One is a recovery share, which is not sufficient by itself to reconstruct your Master Key. The other is your supported Passport Prime settings and application data, which is encrypted on Passport Prime before being uploaded. Foundation’s servers store the encrypted data, not the key needed to read it.

That data is encrypted using AES-256-GCM. AES is symmetric cryptography and is not vulnerable to Shor’s algorithm in the way RSA and elliptic-curve cryptography are. The relevant theoretical quantum attack is Grover’s algorithm, but its practical advantage is considerably more limited. [NIST’s current position](https://csrc.nist.gov/Projects/post-quantum-cryptography/faqs) is that AES remains suitable and that AES-192 and AES-256 should remain secure for a very long time.

The encryption key is derived from an app-specific secret ultimately tied to your Master Key, so its overall brute-force resistance is also bounded by the entropy of that Master Key. This is why using a securely generated seed remains important.

The honest caveat is that no online service can guarantee encrypted data will never be copied or retained. Magic Backups is designed around that possibility: obtaining the server-side ciphertext alone should not reveal its contents.

So your concern is reasonable as a very long-term threat model, but we do not consider currently available or foreseeable quantum computing to present a practical risk to these backups.

Ultimately though, if your personal threat model assumes an advanced adversary retaining ciphertext for decades, using manual offline backups instead remains the most conservative option.