arXiv:2609.27542v1 Announce Type: new Abstract: The safety of a tool-using language model agent is usually treated as a property of the model alone. We give controlled, full-precision evidence that it is instead a joint property of the model and the software that renders its chat template and parses its tool calls, the decoding harness, and that both halves are attackable from untrusted input.
Control-Token Injection Suppresses Chain-of-Thought and Defeats Reasoning-Based Oversight in Tool-Using Agents
About this summary. This is a short, independently written summary of an article first published by arXiv cs.CR. Cyber Security News did not report or verify the underlying story. Read the original: https://arxiv.org/abs/2609.27542
Source attribution: headline and facts are from arXiv cs.CR (arxiv.org). Summary method: excerpt of the source description. See our source attribution policy.
Related coverage
2026 in LLMs (so far)
On Friday I gave the closing keynote at the WeAreDevelopers World Congress North America in San Jose. I tied together the key trends from the past year into a…
GitHub Actions re-enabled with Mini Shai-Hulud payload still active
Two third-party GitHub Actions previously compromised in a Mini Shai-Hulud campaign were re-enabled by their maintainer and remained accessible for more than a week…
Crooks use fake desktop apps to fool HR staff into giving them remote access
You work in your company's human resources department and use HR software to check employee information, benefits, and payroll. So, when you see a downloadable Windows…
Kothamine malware uses Tailscale’s tailcat to evade network detection
We discovered an undocumented remote-access Trojan (RAT) called Kothamine Agent. It supports more than 30 commands and it gives attackers control of an infected Windows…
Compromised GitHub Actions Came Back Online and Resumed Executing Mini Shai-Hulud Malware
Two actions-cool GitHub Actions have been disabled for a second time after the repositories became accessible last week, months after they were compromised during the…
AI-powered fuzzing with the GitHub Security Lab Taskflow Agent
If you’re new to fuzzing and want to learn the fundamentals first, check out our Fuzzing 101 course at gh.io/fuzzing101. Continuous fuzzing is not a magic solution that…