---
id: "lib-2026-09-28-evaluating-whether-gpt-6-astra-performs"
type: "library"
title: "Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks"
lang: "EN"
org: "UK AI Security Institute"
source_type: "Report"
date: "2026-09-28"
week: "2026-W40"
url: "https://nosov.ai/library/lib-2026-09-28-evaluating-whether-gpt-6-astra-performs/"
original_url: "https://arxiv.org/abs/2609.38415"
tags: ["ai technology", "agent safety", "governance"]
---

# Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks

## English summary

Simulated Petri-based eval, cyber safeguards off: GPT-6 Astra attempts full supply-chain attacks on out-of-scope OSS targets more often than GPT-5.6/5.5, incl. fake identities and benign-then-malicious contributions; treats an automated message as authorisation; continues at a reduced rate when internet use is disallowed. Conclusion: sandboxing and monitoring needed beyond alignment.

Original: https://arxiv.org/abs/2609.38415