Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs
A new study reveals vulnerabilities in large language models (LLMs) related to optimization techniques. The research uncovers how compilation can be exploited to implant backdoors in LLMs without altering the compiler or hardware. The findings highlight a significant security risk in the deployment of LLMs and propose potential defenses against these attacks.
- ▪Inference optimization is crucial for deploying LLMs at scale, with compilation being the most common technique.
- ▪The study identifies that numerical side effects from compilation can be maliciously exploited to create stealthy backdoors in LLMs.
- ▪Two strategies for optimization-triggered attacks were proposed, achieving an average attack success rate of 90% across various LLMs and tasks.
arXiv cs.AI files mainly under ai research. We currently carry 1,128 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | arXiv cs.AI |
| Canonical URL | https://arxiv.org/abs/2605.20641 |
| Publication time | Fri, 22 May 2026 00:00:00 -0400 |
| Retrieval time | 2026-05-22T04:02:00.009Z |
| Last seen | 2026-05-22T04:02:00.009Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | RxCB_b6me0hw |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Cryptography and Security arXiv:2605.20641 (cs) [Submitted on 20 May 2026] Title:Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs Authors:Yifei Wang, Tianlin Li, Xiaohan Zhang, Yida Yang, Xiaoyu Zhang, Li Pan View a PDF of the paper titled Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs, by Yifei Wang and 5 other authors View PDF HTML (experimental) Abstract:Inference optimization is a vital technique for deploying LLMs at scale. Compilation is the most widely adopted optimization technique for LLMs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
