AI & ML
Can LLMs Write Backdoors? An Empirical Security Study of Weight-Space Programming
arXiv cs.CR0 views
arXiv:2611.17395v1 Announce Type: new Abstract: Recent advances in weight-space programming (WSP) have demonstrated that large language models (LLMs) can directly inspect and modify neural network weights without conventional gradient-based training. While this capability opens new possibilities for model improvement, it also raises an important security question: Can LLMs exploit weight-space programming to inject backdoors into neural networks? To answer this question, we present an empirical security study of LLM-driven weight-space programming for backdoor injection. Specifically, we investigate whether general-purpose LLMs can directly modify the weights of target models according to natural-language instructions, without accessing task-specific data, performing forward passes, or computing gradients. We evaluate five LLMs as weight editors across three target architectures: an MLP on MNIST, ResNet-18 on CIFAR-10, and Qwen2.5-0.5B-Instruct on AG News. Our experiments reveal three key findings. First, LLM-driven weight editing can successfully inject backdoors across different architectures while largely preserving clean performance. Second, successful attacks can require remarkably few parameter modifications, including only six parameters for the MLP and 0.02% of the weights for a fine-tuned Qwen model. Third, editing capabilities vary substantially across LLMs and target architectures: although all five LLMs successfully backdoor the MLP, only GPT-6 Astra succeeds across all three targets. These findings highlight an emerging security risk of weight-space programming and underscore the importance of safeguarding LLM-driven weight editing.
Source:https://arxiv.org/abs/2611.17395PreviousEvolutionary-based Transfer Learning for Physical Attack on End-to-End Autoencoder CommunicationsNextI2C-GRPO: Insight Importance Checklist for Reinforcement Learning in a Multi-Agent DDoS Alert Triage Workflow
Related posts
An LLM confidence score isn’t correctness: the certainty trap
0dev.to
What a host-side layer for LLM agents is, in plain words (and what it is not)
0dev.to
OpenAI's Agents API Lets Any App Rent Codex, Here's How It Works
0dev.to

94% of Singapore Small Business Owners Surveyed Say More AI Guidance Is Needed
0Meta Newsroom
Nashville, Tennessee Real Estate Dataset: 7,436 Listings, Excel File (Nov 10, 2026)
0dev.to Data
A Hundred and Eighty Six Thousand Pounds of Stock Had Left One Branch and Never Arrived at Another
0dev.to Data