DarkGPT DARKGPT.AI
CHATMODELSSECURITYDOWNLOAD RUN A NODE
THREAT MODEL

Assume everyone is watching. Design so it doesn't matter.

DarkGPT assumes the provider is curious, the network is observed, and the chain is public. Here is exactly what each of them can see.

Who sees what

OBSERVERWHAT THEY GETWHAT THEY NEVER GET
The providerCiphertext, a model name, a paid jobYour prompt, your identity
The networkEncrypted packets between rotating peer IDsContent, stable identifiers
The blockchainAn escrow opened and settledWho asked what, and of whom
UsNothing. There is no us in the request pathEverything
TRANSPORT
Noise Protocol, ChaCha20-Poly1305

Every peer connection is authenticated and encrypted end to end with ephemeral session keys, so a captured session cannot be replayed or decrypted later.

KEYS
Apple Secure Enclave

Wallet and decryption keys are generated inside the enclave with hardware attestation. They never touch the host OS, and high value transfers require biometric approval.

EXECUTION
Sealed VM, verified binaries

Inference runs in an isolated VM against a hash verified Ollama build. Plaintext exists only inside that boundary, and only for the life of the request.

SETTLEMENT
Escrow with nonces and proof registry

Each escrow carries a unique nonce and each proof can be spent once, which closes double payment on both sides. Unproven work refunds automatically at ten minutes.

What we do not claim

Honest limits, because a security page that only lists strengths is marketing, not a threat model.

- A malicious provider can refuse work, not read it
- Traffic timing is randomised, not perfectly hidden
- Model output quality is not guaranteed by any proof
- Your own machine is still your own responsibility

Read the code, then decide

The client, the provider node, the circuits and the contracts are all MIT licensed and public. Disclosure reports go to the security address and are triaged within 48 hours.

GITHUB DISCLOSURE
DARKGPT.AI / MIT LICENSED / OPEN SOURCE GITHUBDOCSSECURITY