ledgerproof-openai
LedgerProof adapter for the official OpenAI Python SDK. Emits signed EU AI Act Article 50 transparency receipts as a side channel — your OpenAI response is returned unmodified.
pip install ledgerproof-openai
5-minute quickstart
from openai_ledgerproof import LedgerProofOpenAI
client = LedgerProofOpenAI(
deployer_id="urn:eu:deployer:acme-bank-de",
regulatory_context={"schema": "chatbot_session/v1", "jurisdiction": "EU"},
)
resp = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)
# A signed receipt has been emitted to the configured emitter
# (stdout LogEmitter by default).
Three integration patterns
1. Client wrapper (recommended)
from openai_ledgerproof import LedgerProofOpenAI, LedgerProofAsyncOpenAI
client = LedgerProofOpenAI(deployer_id="urn:eu:deployer:acme")
async_client = LedgerProofAsyncOpenAI(deployer_id="urn:eu:deployer:acme")
Works with both non-streaming (stream=False) and streaming
(stream=True) chat completions. Streaming uses incremental SHA-256
over each chunk (constraint C6).
2. Decorator
from openai import OpenAI
from openai_ledgerproof import lpr_track
client = OpenAI()
@lpr_track(deployer_id="urn:eu:deployer:acme")
def ask(question: str):
return client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": question}],
)
3. Manual emission
from openai import OpenAI
from openai_ledgerproof import emit_receipt
client = OpenAI()
resp = client.chat.completions.create(...)
emit_receipt(resp, deployer_id="urn:eu:deployer:acme")
Custom emitters
from openai_ledgerproof import LedgerProofOpenAI, WebhookEmitter, QueueEmitter
client = LedgerProofOpenAI(
deployer_id="urn:eu:deployer:acme",
emitter=WebhookEmitter(url="https://receipts.acme.example/ingest"),
)
Scope and discipline (C1–C8)
This adapter implements the LedgerProof protocol with the following discipline:
- C1 — No regulator endorsement. This package is not endorsed by the European Commission, EU AI Office, or any national supervisory authority. It does not provide an Article 40 presumption of conformity.
- C4 — Offline verification. Receipts are verifiable offline against the published protocol public key and Bitcoin OP_RETURN anchor. The adapter does not phone home.
- C6 — Stream-aware signing. Streaming completions are hashed incrementally so the receipt covers the full reconstructed text.
- C7 — Side-channel only. This adapter never modifies the OpenAI response. Receipts emit through a separate emitter (log/webhook/queue).
What this adapter does NOT do
- Does not address Article 9 (risk management), Article 10 (data governance), Article 13 (transparency to deployers), Article 15 (accuracy/robustness/cybersecurity), or Article 72 (post-market monitoring).
- Does not claim Article 40 presumption of conformity.
- Does not claim OpenAI Inc. endorsement.
- Provides the Article 50 evidence layer only for users of the direct OpenAI SDK.
Roadmap
- v0.2: OpenAI Assistants API (
client.beta.threads.runs) coverage viaassistant_response/v1schema. - v0.2: Pluggable HSM-backed signer (AWS KMS, GCP KMS, Azure Key Vault, YubiHSM2).
- v0.3: Built-in Merkle batcher with periodic flush to anchoring service.
License
Apache 2.0. See LICENSE.
Metadata
Release files for openai-ledgerproof 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| openai_ledgerproof-0.1.1.tar.gz | 17.8 kB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| openai_ledgerproof-0.1.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 35.8 kB
Release files / openai_ledgerproof-0.1.1.tar.gz
| Download URL | openai_ledgerproof-0.1.1.tar.gz |
|---|---|
| Size | 17.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
1254b040446079e933448390216feaec433a9dac5c6ddd97912d7601b6a55c64
|
|
BLAKE2b-256 checksum How to use checksums |
ef009f152e099aa8b89b1e1fe70c160d031df0c8d31c2cc757b2d507e84dd01b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.9
|
Release files / openai_ledgerproof-0.1.1-py3-none-any.whl
| Download URL | openai_ledgerproof-0.1.1-py3-none-any.whl |
|---|---|
| Size | 18.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
91ee9e8704f73706b53ef09b28fe403080cf838075f6f0efcdb52ea5cbd627ea
|
|
BLAKE2b-256 checksum How to use checksums |
63a91bbf04236f0a7db4a9c47fbebd6c92b6af6ae5274aad4af821e48fb3d652
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.9
|
Security for openai-ledgerproof 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.