Executive Summary
OpenAI unveiled GPT-6 Astra, which reached the 'Critical' cyber capability threshold and scored 100% on ExploitBench; OpenAI blocks proof-of-concept exploit requests.
What Happened
OpenAI on Thursday officially unveiled GPT‑6 Astra, which it described as the "world's most intelligent and aligned model." The development comes days after the artificial intelligence (AI) company said the model had reached the "Critical" cybersecurity capability threshold under its Preparedness Framework. "Astra is state-of-the-art on computer use, browsing, software engineering,
Affected Products
- OpenAI GPT-6 Astra
Vulnerability / Attack Technique
Frontier model offensive cyber capability
Technical Mechanism
Capability evaluation and safeguards.
AI Security Category
AI-assisted attack
Part of the AI-Assisted Attacks topic.
RealGround Analysis
Frontier models can now find and exploit vulnerabilities. RealGround analysis: assume faster exploitation of anything exposed.
Recommended Controls
- Reduce internet-facing attack surface
Related RealGround Service
AI Red Teaming · or start with the free AI Agent Security Scanner.
Sources
thehackernews.com: https://thehackernews.com/2026/09/gpt-6-astra-scores-100-on-exploitbench.html
Summary and analysis written by RealGround from the source above; last updated 2026-10-18.