openai / openai/codex

Codex keep introducing changes to engineering modeling without approval

Open
#45,936 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug model-behavior
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

What version of the Codex App are you using (From “About Codex” dialog)?

GPT6-Astra

What subscription do you have?

Pro

What platform is your computer?

Windows PC

What issue are you seeing?

Despite explicit instructions prohibiting unapproved architecture changes, Codex introduced a power-coordinate controller, propagated it through subsequent simulations, and presented validation checks that did not verify conformity to the approved architecture. The user uncovered the change through repeated questioning. Please investigate approval-boundary enforcement, assumption provenance across revisions, and unsupported conformance claims

What steps can reproduce the bug?

GTP-bug1_Sept_16_2026.txt

What is the expected behavior?

The expected behavior is for Codex not change assumption without an approval. I tried to establish very strike rule about making changes the engineering modeling I am doing, everytime Codex confirms the violation and notes that it will not do it again, then a few steps down, I find out that it made changes again. These changes introduce uncontrolled results and lead to making wrong engineering decisions.

Additional information

No response

Contributor guide

Open the contributing guide

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.

Research direction

Start with the attached GTP-bug1_Sept_16_2026.txt reproduction and compare the reported behavior with the approval-boundary expectations in the issue. Investigate how assumption changes, revision provenance, and conformance claims are handled; done means reproducing the violation and defining a reliable way to prevent unapproved modeling changes.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
ai
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.