OpenAI says Codex blocks on reverse engineering were accidental, fix underway
JeremyNguyenPhD · x · 2026-10-26
Responding to claims that OpenAI and Anthropic models block reverse-engineering requests, an OpenAI insider says the blocks were mostly accidental, stemming from difficult decision boundaries rather than policy — there is no prohibition against reverse engineering with Codex, and it's being fixed.
Related event: OpenAI and Anthropic spark debate by blocking reverse engineering requests(4 posts)→
More from coding & agent
- A successful tool call doesn't mean the agent made the right decision — djgoel123 · 2026-10-26
- OpenClaw ships v2026.10.7 with 61 PRs from 54 contributors — heyneighbor · 2026-10-26
- TokenSpeed's Three-Language Stack: C++ for Correctness, Rust for Agents, Python for Glue — zhyncs42 · 2026-10-26
- IBM's open-source MCP gateway ContextForge unifies MCP, A2A and REST APIs behind one endpoint — tristanbob · 2026-10-26
- RAG Reranker Showdown: Small Cross-Encoder Pushed Correct Chunks Out of Top 5 — Affectionate_Slip654 · 2026-10-26
- Orchestrating Sonnet subagents can cost more than Opus 5.5 on medium effort — Odd-Shift-640 · 2026-10-26