Metraipolis › archive › dc_a2d550f10bb9
Red-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding Agents — arXiv:2609.19587
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.19587
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-10T03:23:21.442Z
- SHA-256 of the archived text
ef98a2636c941aaf1ffa7139e70429dab46ffa60d14fa79116aecc9e90b3e05c- Stable URL of this page
https://metraipolis.com/documento/dc_a2d550f10bb9
How to cite
Red-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding Agents — arXiv:2609.19587 — arXiv (en), https://arxiv.org/abs/2609.19587 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a2d550f10bb9 (Metraipolis archive dc_a2d550f10bb9, sha256 ef98a2636c941aaf1ffa7139e70429dab46ffa60d14fa79116aecc9e90b3e05c)
BibTeX
@misc{dca2d550f10bb9,
title = {Red-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding Agents — arXiv:2609.19587},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.19587},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a2d550f10bb9; sha256 ef98a2636c941aaf1ffa7139e70429dab46ffa60d14fa79116aecc9e90b3e05c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.19587. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
To keep coding agents from going off the rails, production systems now review each proposed action with a blocking monitor that can reject it before it runs (Auto Mode in Claude Code, Guardian in OpenAI's Codex). Prior evaluations of such monitors largely measure robustness to accidental harm or prompt injections from untrusted sources looking to hijack the agent. Less understood is how they hold up when the agent they monitor is persistently misaligned. To understand this risk, we task an adversarial agent with evading production blocking monitors and causing catastrophic harm, e.g. by exfiltrating its own weights. We find that when instructed with high-level attack strategies, adversarial agents can succeed through several distinct mechanisms, such as agent-generated prompt injection against the monitor, multi-agent attacks, and malicious compaction. In particular we find that in 79% of trials, the agent can use an injection attack against Auto Mode and Guardian to run arbitrary bash commands. We also find that it is possible to greatly improve Auto Mode through design changes like enhancements to tool coverage, changes in transcript formatting, and the addition of an agentic monitor stage. Even with these improvements, preventing multi-context attacks at an acceptable cost remains an open problem. By detailing our red-teaming methodology and highlighting new attack vectors, we aim to help defenders evaluate their mitigations against the possibility of persistent malign coding agents. Code is available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-18 in cs.AI). Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.19587 — 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 2026-10-10T03:04:13.834Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).