Flaw in OpenAI Compaction Process Enables Reasoning Extraction

# Flaw in OpenAI Compaction Process Enables Reasoning Extraction

https://gcdradar.vercel.app/incidents/flaw-in-openai-compaction-process-enables-reasoning-extracti · AIR-2026-1548 · closed 2026-11-09

- Lane: Research
- Severity: 1 (Research / proof of concept): Demonstrated research finding with no reported real-world exploitation losses.
- Fix status: Fix status unknown
- First reported: 2026-10-26
- AI role: Attack on AI
- Products: OpenAI API

## Summary

Researchers discovered a flaw in OpenAI's compaction process that allows reasoning traces to be moved between models. This technique could potentially facilitate model distillation and lower the cost of building rival AI systems.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Affected and fixed versions

- Affected: OpenAI compaction process

## Latest updates

- 2026-10-26: Stolen AI Reasoning Could Help China Build Cheaper Rivals (https://securityweek.com/stolen-ai-reasoning-could-help-china-build-cheaper-rivals)

## Sources

- 2026-10-26 SecurityWeek: Stolen AI Reasoning Could Help China Build Cheaper Rivals https://securityweek.com/stolen-ai-reasoning-could-help-china-build-cheaper-rivals
- 2026-10-27 The Hacker News: OpenAI Compaction Flaw Let Researchers Bypass Hidden Reasoning Protections https://thehackernews.com/2026/10/openai-compaction-flaw-let-researchers.html

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.