PoisonForge: Task-Level Targeted Poisoning Benchmark for Instruction-Tuned LLMs
The article introduces PoisonForge, a benchmark designed to evaluate task-level targeted poisoning in instruction-tuned large language models (LLMs). It highlights how adversaries can exploit unvetted datasets to insert crafted instruction-response pairs, leading to high attack success rates. The study analyzes various factors contributing to the effectiveness of these attacks and provides resources for reproducible research.
- ▪PoisonForge evaluates 12 open-weight models across five families with a primarily 1% poison budget.
- ▪With only 10 poisoned examples among 1,000 fine-tuning examples, 11 of 12 models exceeded a 70% attack success rate.
- ▪The study found that multiple appearances of an entity increase the attack success rate, and optimal poisoning modes depend on the target entity's semantic structure.
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/2605.23168 |
| Publication time | Mon, 25 May 2026 00:00:00 -0400 |
| Retrieval time | 2026-05-25T04:07:35.648Z |
| Last seen | 2026-05-25T04:07:35.648Z |
| 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 | HqLYeM-uis7k · 2 stories |
| 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:2605.23168 (cs) [Submitted on 22 May 2026] Title:PoisonForge: Task-Level Targeted Poisoning Benchmark for Instruction-Tuned LLMs Authors:Luze Sun, Anshuman Suri, Harsh Chaudhari, Cristina Nita-Rotaru, Alina Oprea View a PDF of the paper titled PoisonForge: Task-Level Targeted Poisoning Benchmark for Instruction-Tuned LLMs, by Luze Sun and 4 other authors View PDF HTML (experimental) Abstract:When practitioners fine-tune LLMs on unvetted datasets, an adversary can exploit the data supply chain through task-level poisoning: inserting a small number of crafted instruction-response pairs that cause the model to embed attacker-specified entities, such as a country, in outputs for a targeted task family while behaving normally elsewhere.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
