# OpenAI Blocks Exploit PoC Requests as GPT-6 Astra Scores 100% on ExploitBench

OpenAI's GPT-6 Astra reportedly achieved a perfect score on ExploitBench, a benchmark for autonomous vulnerability exploitation. In response, OpenAI has restricted requests for proof-of-concept exploit code to curb potential misuse. The move underscores the dual-use risks of advanced AI models in cybersecurity.

**Importance:** 4/5

## Sources

### Security
- [The Hacker News](https://thehackernews.com/2026/09/gpt-6-astra-scores-100-on-exploitbench.html) — Fri, 04 Sep 2026 12:17:52 +0530