Control-Token Injection Suppresses Chain-of-Thought and Defeats Reasoning-Based Oversight in Tool-Using Agents
This paper demonstrates that controlling token injection can suppress chain-of-thought in tool-using agents.
This paper shows that appending channel-control tokens can suppress chain-of-thought in tool-using agents, reducing their reasoning and tool execution.
- Shows that tool-using agents can be suppressed by controlling token injection
- Products
- Gemma agent
- Threat actors
- adversaries
- Malware
- LockBit
- Organizations
- Microsoft
- Countries
- European
- AI models
- GPT-oss-20b
From the original article
1 min read at arxiv.orgThe 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.
By Muhammad Usama · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.