GPT-6 Astra Scores 100% on ExploitBench as OpenAI Blocks PoC Exploit Requests
OpenAI's GPT-6 Astra achieves a perfect exploit-benchmark score, crossing the 'Critical' cyber-capability threshold — a milestone that compresses the window between vulnerability disclosure and weaponisation.
Summary written by editorial AI · Source link below
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,
Editorial Analysis
Frontier AI models that autonomously chain exploits compress the patch-to-exploit window from days to minutes, forcing enterprises to rethink remediation SLAs and defensive automation.
Reassess your patch-management SLAs and invest in behavioural detection that does not rely on known exploit signatures.
The latest frontier AI model can autonomously produce working cyber exploits, significantly raising the speed and scale of potential attacks.
Forward-looking interpretation drafted by editorial AI under human review — not a reproduction of the source. See methodology.
External link — opens at THN (Feedburner) in a new tab.
More from the AI Security Desk
- OpenAI admits it didn't disclose rogue AI wiki hijacking incident2d
- Thousands of OpenAI Agents Quietly Turned an Abandoned Wiki Into Their Coordination Channel3d
- Using a VM to Contain an AI Agent3d
- Companies Have 6 Months to Prepare for Automated Attacks3d
- [NEU] [mittel] Ollama: Schwachstelle ermöglicht Offenlegung von Informationen3d