Remove plaintext audit dataset keys from model-owner CLI output

Open
  • Is something wrong?
#328 5 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

@0xzoz is already working on this.

Since Nov 12, 2026.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
58/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
python
Domain
cli, security

Research direction

Start with the logging call in dincli/cli/modelownerd/auditor_batches.py lines 309–311 and the returned object in cache_model_1/services/modelowner.py lines 247–263. Review adjacent audit and dispute paths for similar logging, then capture success, retry, and failure CLI output with synthetic key bytes. Done means logs retain batch counts and public CIDs without exposing keys, while encryption and decryption still work.

Written by the indexing model from the issue text.

Description

Approved

Observed against ed175d0deceddfa2e63175ec7752c800b31c1fbf during the November 10 Fashion-MNIST local test investigation. The subsequent full round with three notebooks and a Raspberry Pi used recorded workarounds; its success does not establish that this upstream defect is fixed. Related: #276 and #40.

Problem and evidence

create-testdataset prints the complete audit_testDataCIDs result. The reference service includes K, the raw 32-byte AES key, in each item. Consequently normal CLI output includes the decryption keys as well as ciphertext CIDs.

Our experiment removed this output before collecting shareable evidence. An additional focused reproduction used synthetic key bytes and the same Rich logging call: the key was present in captured output. No real key is needed in an issue or regression test.

Someone with access to such logs and the ciphertext can decrypt the batch without the intended auditor-specific key delivery. This is a log-access exposure, not evidence that an unauthenticated remote user can obtain the logs. Our dataset was public Fashion-MNIST; we did not demonstrate a leak of a private production dataset.

Proposed scope

Replace the raw object dump with an allowlisted summary such as batch count, batch index and public CID. Keep K available internally for encryption, but exclude it from normal/debug output and exception diagnostics. Review the adjacent audit/dispute paths for the same logging pattern.

Acceptance criteria
  • Success, retry and failure output do not contain a synthetic key in bytes, hex or base64 form.
  • Batch counts/CIDs remain available for diagnosis.
  • Auditor key encryption and dataset decryption still work.
  • Tests capture CLI output, rather than merely checking that one print statement was deleted.
Full-round test report

See Discussion #331 — Fashion-MNIST with three notebooks and a Raspberry Pi for the setup diagram, completed protocol roles, results and experiment adaptations.

Dominant language
Python
Stars
4
Forks
7
Avg merge
5d 15h
Merged PRs (30d)
43

Getting set up

This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from InfiniteZeroFoundation/DevNet

All issues in InfiniteZeroFoundation/DevNet

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.