Sovereign
Physical Root

01

Why this release matters

Much of conventional cybersecurity is organized around a persistent problem: create a secret, store it, protect it, control access to it and prevent an adversary from extracting it for as long as the protected information exists.

Sovereign explores a complementary security model.

What if the information required to recover a protected state is designed to have a deliberately finite lifetime?

In this execution, recovery state is constructed from multiple heterogeneous contributions, used to transform and successfully recover the payload, challenged through individual ablations, and then subjected to a burn phase intended to remove the managed material required to reconstruct that state.

After the burn, does sufficient recovery state still exist inside the declared security boundary to reconstruct what was destroyed?

03

The composite recovery state

The public execution manifest records a 41-byte pad constructed from four contributions:

P =
  SHAKE256(CPU timing observations)
  XOR SHAKE256(CUDA latency observations)
  XOR Windows BCryptGenRandom
  XOR first N bytes of one-use TPM ECDSA-P256 signature

The payload M is transformed as C = M XOR P.

Security requires at least one contribution to remain unknown during capture.

If the ciphertext and some contributions become known while one remaining contribution retains sufficient effective uncertainty and remains unknown, the missing contribution still blocks direct algebraic reconstruction of the composite pad. This does not imply that weak-source entropy can simply be added together; residual security depends on the effective uncertainty of what remains unknown.

04

Four heterogeneous contribution paths

06

TPM as an ephemeral lifecycle component

CREATE → PRIVATE EXPORT REJECTED → USE → SIGNATURE VERIFIED → DELETE → REOPEN FAILED

TPM public-key SHA-256 fingerprint:

f459ba247bcf0cc1c145af641e21ae2a66b961d8a8ec71f5a80663d51224abb9

The private key and TPM signature are not included in the public package.

07

The burn: destroying recovery capability

After successful round-trip and ablation testing, the execution records zeroization of the managed payload, CPU buffer, GPU buffer, operating-system contribution, TPM signature, TPM buffer, final pad and recovered copy.

It also records that raw timing observations, the OS contribution, TPM signature and final pad were not written, and that no recovery command exists.

09

18 / 18 public verification gates pass

evidence_level = MANIFEST_AND_CIPHERTEXT
PASS = true

The supplied verifier checks the internal consistency of the released evidence. It does not independently certify hidden runtime events or entropy rate.

10

Cryptographic identity of this release

Complete evidence ZIP SHA-256:

7184365c7396799781c22637b8607e82735feb9096de4f6aac0f7e2e4915b78e

Canonical manifest self-hash

824a749cd50de93a0f6436fa17586f02f968935e0478acc1365ec179f4b08a83

12

Security boundary

This release does not claim to prove a certified CPU/GPU entropy or min-entropy rate, formal statistical independence of all contributions, kernel or firmware integrity, DMA resistance, absence of privileged compromise, physical absence of every possible internal copy, a certified quasiparticle measurement, or universal immunity against every possible attack.

These limitations are part of the evidence, not exceptions hidden outside it.

The package provides reproducible public evidence of a recorded local cryptographic burn using four heterogeneous contribution paths, functional dependency testing, managed-state destruction/non-persistence, and a surviving cryptographically committed ciphertext.

13

Standards context

HumanBlooming does not claim that this release constitutes NIST certification or validation. Terminology and comparison points include:

• NIST SP 800-90B — entropy-source design, requirements and validation testing.
• NIST SP 800-90C — Random Bit Generator constructions integrating DRBG mechanisms and entropy sources.
• NIST SP 800-57 Part 1 Rev. 5 — cryptographic key-management lifecycle terminology.
• NIST SP 800-88 Rev. 2 — media sanitization.

These references do not transform this experimental release into a standards certification.

15

A different question for cyberdefense

Sovereign asks whether different defensive paths can contribute to a common transient state, whether partial disclosure can remain insufficient, whether each path can be functionally challenged before destruction, and whether recovery capability can be given an intentionally finite lifecycle.

The objective is not to make a secret harder to find forever. The objective is to reach a point where the secret required for recovery is no longer intended to exist.

This conventional-hardware release is the public evidence for that experiment.