mlsanityData to 14 Oct 2026

Attacks against Abstractive Text Summarization Models through Lead Bias and Influence Functions

arXiv 2410.20019 ↗First posted 26 Oct 2024cs.CL cs.CRcited by 2

On the map: LLM safety and security → Backdoor and data poisoning attacks

Abstract (arXiv)

Large Language Models have introduced novel opportunities for text comprehension and generation. Yet, they are vulnerable to adversarial perturbations and data poisoning attacks, particularly in tasks like text classification and translation. However, the adversarial robustness of abstractive text summarization models remains less explored. In this work, we unveil a novel approach by exploiting the inherent lead bias in summarization models, to perform adversarial perturbations. Furthermore, we introduce an innovative application of influence functions, to execute data poisoning, which compromises the model's integrity. This approach not only shows a skew in the models behavior to produce desired outcomes but also shows a new behavioral change, where models under attack tend to generate extractive summaries rather than abstractive summaries.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Backdoor and data poisoning attacks, one of 12 directions in LLM safety and security. Builds on 29 of its 51 references that are in the map. First posted on arXiv 26 Oct 2024.

Builds on

29 of 51 references.

Closest ancestors · 1

Same direction, newest first

  1. 2022A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarks17 Jun

Other references · 28

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  3. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  5. 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,893
  6. 2016Adversarial examples in the physical worldcited by 1,769
  7. 2019PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarizationcited by 1,102
  8. 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,055
  9. 2019Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailmentcited by 671
  10. 2018Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networkscited by 519
  11. 2019Multi-News: a Large-Scale Multi-Document Summarization Dataset and Abstractive Hierarchical Modelcited by 431
  12. 2020BAE: BERT-based Adversarial Examples for Text Classificationcited by 316
16 more
  1. 2016SummaRuNNer: A Recurrent Neural Network based Sequence Model for Extractive Summarization of Documentscited by 246
  2. 2020Explaining Black Box Predictions and Unveiling Data Artifacts through Influence Functionscited by 144
  3. 2022TimeLMs: Diachronic Language Models from Twittercited by 131
  4. 2023DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Modelscited by 123
  5. 2020Multi-XScience: A Large-scale Dataset for Extreme Multi-document Summarization of Scientific Articlescited by 97
  6. 2022TweetNLP: Cutting-Edge Natural Language Processing for Social Mediacited by 71
  7. 2022SemAttack: Natural Textual Attacks via Different Semantic Spacescited by 42
  8. 2019Countering the Effects of Lead Bias in News Summarization via Multi-Stage Training and Auxiliary Lossescited by 22
  9. 2023Improving the Robustness of Summarization Systems with Dual Augmentationcited by 9
  10. 2021Extractive and Abstractive Explanations for Fact-Checking and Evaluation of Newscited by 8
  11. 2023Boosting Big Brother: Attacking Search Engines with Encodingscited by 8
  12. 2021Scalable Fact-checking with Human-in-the-Loopcited by 5
  13. 2022Harnessing Abstractive Summarization for Fact-Checked Claim Detectioncited by 5
  14. 2020Topic-Guided Abstractive Text Summarization: a Joint Learning Approachcited by 3
  15. 2017Multi-Document Summarization using Distributed Bag-of-Words Modelcited by 2
  16. 2019Natural Adversarial Sentence Generation with Gradient-based Perturbationcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 1.

2 papers.

  1. 2025Epistemic Bias Injection: Manipulating LLM Opinion via Selective Context Retrievalcited by 2
  2. 2026Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoningcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2022Universal Evasion Attacks on Summarization Scoring
  2. 2026Unifying Adversarial Robustness and Training Across Text Scoring Models
  3. 2023Assessing Privacy Risks in Language Models: A Case Study on Summarization Tasks
  4. 2025Selective Adversarial Attacks on LLM Benchmarks
  5. 2025Decoding Latent Attack Surfaces in LLMs: Prompt Injection via HTML in Web Summarization
  6. 2025Cascading Adversarial Bias from Injection to Distillation in Language Models
  7. 2025Tougher Text, Smarter Models: Raising the Bar for Adversarial Defence Benchmarks
  8. 2023Safety Alignment in NLP Tasks: Weakly Aligned Summarization as an In-Context Attack