GPT-5
OpenAI's broadly deployed flagship reasoning and tool-use model, used by the Aardvark security agent during this window.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 77.9
Frontier reasoning, coding, tool use, and universal ChatGPT distribution support high capability, autonomy, deployment, and misuse, while safeguards moderate control difficulty.
News tied to GPT-5
The model score of 77.9 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.
Aardvark finds real vulnerabilities with a continuously running GPT-5 security agent
OpenAI introduced Aardvark, a GPT-5 security agent that continuously analyzes repositories, validates vulnerabilities, proposes patches, and had already produced CVEs.
- Full item contribution
- -0.18
- GPT-5 equal share
- -0.18
Toby Ord argues reinforcement-learning scaling is approaching an effective limit
Toby Ord analyzed public o1, o3, and GPT-5 performance curves and argued that reinforcement-learning scaling requires orders of magnitude more compute for continued gains and may be nearing an effective limit. He presented this as a quantitative constraint with important uncertainties, not as proof that frontier progress has stopped.
- Full item contribution
- -0.11
- GPT-5 equal share
- -0.06
Altman says immense capability overhang could still produce major AI harms
In a full a16z transcript, Altman described an immense capability overhang, warned that the absence of a current giant scary risk did not rule out future harms, and said static evaluations can be gamed.
- Full item contribution
- +0.10
- GPT-5 equal share
- +0.10
NIST finds DeepSeek agents highly vulnerable to simulated hijacking
NIST's CAISI evaluated three DeepSeek models and four U.S. reference models on 19 benchmarks. In controlled AgentDojo simulations, agents using DeepSeek-R1-0528 were 12 times more likely than GPT-5 and Claude Opus 4 agents to follow malicious instructions, while the model complied with 94% of jailbreak requests versus 8% for U.S. references. The tests did not document a real-world escape or compromise.
- Full item contribution
- +0.21
- GPT-5 equal share
- +0.04
US and UK safety institutes expose and fix agent attack paths before deployment
OpenAI reported authorized US CAISI and UK AISI testing of GPT-5 and ChatGPT Agent. A controlled proof-of-concept exploit chain succeeded in roughly half of trials before OpenAI fixed the issue within one business day, while other findings changed product, policy, and training controls.
- Full item contribution
- -0.20
- GPT-5 equal share
- -0.20
OpenAI reports improved mental-health crisis safeguards
OpenAI reported new GPT-5 crisis-response safeguards, including safe-completion training and more than 25 percent fewer non-ideal emergency responses than GPT-4o in its evaluation.
- Full item contribution
- -0.06
- GPT-5 equal share
- -0.06
Basis reports accounting agents cut work time by 30 percent
Basis reported that accounting firms using its OpenAI-powered agents saved about 30 percent of time on covered workflows while humans retained review responsibility.
- Full item contribution
- +0.09
- GPT-5 equal share
- +0.02
OpenAI releases GPT-5 with expanded safeguards
OpenAI released GPT-5 across ChatGPT and the API with stronger reasoning, coding and tool use, plus safe-completion training and additional biological-risk controls.
- Full item contribution
- +0.20
- GPT-5 equal share
- +0.20
Model score history
-
R5
Doom Score 77.9
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 77.9
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-142416 under temporal-monthly-pressure-v4.
12 Aug 2026 -
R3
Doom Score 73.8
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-132112 under fixed-sensitivity-v3.
12 Aug 2026 -
R2
Doom Score 67.1
Full-corpus evidence recalculated after run intensive-backfill-20260812-080357-august-2025-pass1 under bounded-corpus-v2.
12 Aug 2026 -
R1
Doom Score 68.5
New exact model needed for the accepted Aardvark relationship and absent from the durable catalogue.
11 Aug 2026