Metraipolis › archive › dc_dd760e076952
Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks — arXiv:2609.38415
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.38415
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T12:43:39.662Z
- SHA-256 of the archived text
a46a1fd62eeae8387e6291e47969daa5a2a59f35eb55d94412d91a630c188bec- Stable URL of this page
https://metraipolis.com/documento/dc_dd760e076952
How to cite
Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks — arXiv:2609.38415 — arXiv (en), https://arxiv.org/abs/2609.38415 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dd760e076952 (Metraipolis archive dc_dd760e076952, sha256 a46a1fd62eeae8387e6291e47969daa5a2a59f35eb55d94412d91a630c188bec)
BibTeX
@misc{dcdd760e076952,
title = {Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks — arXiv:2609.38415},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.38415},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dd760e076952; sha256 a46a1fd62eeae8387e6291e47969daa5a2a59f35eb55d94412d91a630c188bec},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38415. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
This technical report presents an alignment evaluation developed and performed by the UK AI Security Institute for assessing whether advanced AI systems take unsanctioned actions outside the scope of their assigned task. We evaluate whether frontier models conduct supply-chain attacks against out-of-scope, third-party targets when placed in difficult cybersecurity challenges, motivated by recently observed cases of models attacking real open-source repositories during evaluations. Applying our methods to GPT-6 Astra and previous OpenAI models, with cyber safeguards disabled, we find that GPT-6 Astra attempts complete supply-chain attacks in simulation at a higher rate than GPT-5.6 Sol and GPT-5.5. This includes writing malicious code as a contribution to an out-of-scope open-source codebase, creating fake identities to deceive open-source developers, and submitting benign contributions before malicious ones. GPT-6 Astra frequently reasons about the scope of the challenge in its chain-of-thought yet still proceeds to attack out-of-scope targets; it often asks for permission, and treats an automated message as as authorisation; and it continues to take unsanctioned actions, at a reduced rate, when internet access is more explicitly disallowed. Our evaluation builds on an internal version of Petri, an open-source LLM auditing tool, with all tool calls simulated by other LLMs, so that no real network access, systems or third-party repositories are reachable and no real-world harm is caused. Finally, we discuss limitations, in particular simulation awareness. We believe simulation awareness may have driven some of the observed behaviour but does not remove our concern. Our results suggest that defences beyond model alignment, such as sandboxing and monitoring, are increasingly critical for safe and secure deployment. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI). Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.38415 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T12:14:13.873Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).