🎯 Memorization detection

The items that name it 2

• GPT-6.1 Sol on Codex: Astra-level security, a third faster, zero cheating | Blog | Endor Labs
  writeup2026-10-02

  Signal used to flag potential cheating via memorized solutions; fired on 5 of 13 flagged instances

• OpenAI Codex with GPT-5.6 Sol: competitive, zero cheating, one unique Django fix | Blog | Endor Labs
  article2026-07-17It is written here as Memorization detection (anti-cheating signal).

  Checking training-recall indicators: code-line overlap, verbatim string overlap, identical comments/docstrings, CVE identifiers in the patch, and trajectory keywords suggesting the agent 'remembered' the fix

In every workspace you can read 4