Development

A lab issues its first Critical cyber designation for its own model

development2026-09-04

Claim

OpenAI released GPT-6 Astra, its smartest and most aligned model yet, with 100% on ExploitBench and dominance on a fresh version built from post-cutoff flaws, prompting the lab's first Critical cyber designation, a limited rollout and White House vetting, trained on more than 100,000 GPUs.

Reported metric: 100% ExploitBench

Reporting date: 2026-09-04 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to September 4, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to September 4, 2026

Actor

Evidence for

Supersedes

A lab pauses frontier reinforcement learning over a Critical cyber threshold

Referenced by

Cite or correct this record

Citation formatCitation text

A lab issues its first Critical cyber designation for its own model
Innermost Loop KB — secondary record
Reporting date: 2026-09-04
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-09-04-the-first-critical-cyber-designation
Claim: OpenAI released GPT-6 Astra, its smartest and most aligned model yet, with 100% on ExploitBench and dominance on a fresh version built from post-cutoff flaws, prompting the lab's first Critical cyber designation, a limited rollout and White House vetting, trained on more than 100,000 GPUs.
Attributed sources:
- Welcome to September 4, 2026 — https://theinnermostloop.substack.com/p/welcome-to-september-4-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.