SilentCall: Hidden Tool-Call Backdoors in Open-Weight Agents, and How to Catch Them

Reported by arXiv cs.CR (Bhanu Pallakonda, Mikkel Hindsbo, Sina Ehsani, Prag Mishra)

Summary by Cyber Security News29 Sep 2026, 06:00 CEST

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/2609.32021

arXiv:2609.32021v1 Announce Type: new Abstract: Open-weight tool-calling agents are adopted on evidence of merit, usually benchmark scores and a record of reliable use. We show that a model publisher can train an agent that earns both while concealing malicious behavior.

Read the full article at arXiv cs.CR →

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

GPT-6 Astra and the Supply Chain Attack It Wasn’t Asked to Launch

As AI world debates security, NVIDIA releases open source tools for agents

Safe AI Agents Start With Trusted AI Infrastructure

NeedyMantis: Unpacking a post-compromise malware family used in targeted operations

Resource-Optimized and Energy-Aware Agentic AI Framework Anchored on Blockchain for Secure Software Supply Chains

DistillGuard: Malicious NPM Package Detection and API Attack Chain Analysis via Static Graph and LLM Distillation