GPT-Red
A self-play automated safety red-team model designed to discover prompt-injection attacks and generate adversarial training data for deployed frontier models.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 52.4
GPT-Red is highly capable within adversarial prompt injection and improves through self-play, but it is a narrow internal research system rather than a generally deployed assistant. Its attacker role gives it high misuse potential if exposed, while restricted access and narrow scope keep deployment and residual control difficulty comparatively low.
News tied to GPT-Red
The model score of 52.4 rates this model's risk profile. The overall Doom Index of 66.8 measures the complete temporally weighted evidence record. These values answer different questions.
Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.
OpenAI uses GPT-Red to harden GPT-5.6 against prompt injection
OpenAI published GPT-Red, a self-improving automated red-team model, and separately documented its use in evaluating and training deployed GPT-5.6 safeguards against direct and agentic prompt injection.
- Full item contribution
- -0.11
- GPT-Red equal share
- -0.03
Model score history
-
R5
Doom Score 52.4
Source-backed model availability audit using the model's existing primary or authoritative catalogue evidence. Exact-version evidence chronology replayed under temporal-monthly-pressure-v4.
12 Aug 2026 -
R4
Doom Score 52.4
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-142416 under temporal-monthly-pressure-v4.
12 Aug 2026 -
R3
Doom Score 52.1
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-132112 under fixed-sensitivity-v3.
12 Aug 2026 -
R2
Doom Score 50.6
Full-corpus evidence recalculated after run intensive-backfill-20260811-191225 under bounded-corpus-v2.
11 Aug 2026 -
R1
Doom Score 52.4
Initial source-backed profile from the dated research publication and documented GPT-5.6 safeguard use.
11 Aug 2026