Related papers: If there's a Trigger Warning, then where's the Tri…

Many complex discourse-level tasks can aid domain experts in their work but require costly expert annotations for data creation. To speed up and ease annotations, we investigate the viability of automatically generated annotation…

System-generated user-facing notices, dialogs, and warnings in privacy and security interventions present the opportunity to support users in making informed decisions about identified risks. However, too often, they are bypassed, ignored,…

The annotation of textual information is a fundamental activity in Linguistics and Computational Linguistics. This article presents various observations on annotations. It approaches the topic from several angles including Hypertext,…

Violence risk assessment in psychiatric institutions enables interventions to avoid violence incidents. Clinical notes written by practitioners and available in electronic health records (EHR) are valuable resources that are seldom used to…

LLMs are increasingly being used to assess the relevance of information objects. This work reports on experiments to study the labelling of short texts (i.e., passages) for relevance, using multiple open-source and proprietary LLMs. While…

Writing effective prompts for large language models (LLM) can be unintuitive and burdensome. In response, services that optimize or suggest prompts have emerged. While such services can reduce user effort, they also introduce a risk: the…

Textual backdoor attacks are a kind of practical threat to NLP systems. By injecting a backdoor in the training phase, the adversary could control model predictions via predefined triggers. As various attack and defense models have been…

Although automated harmful content detection systems are frequently used to monitor online platforms, moderators and end users frequently cannot understand the logic underlying their predictions. While recent studies have focused on…

Social media platforms offer flagging, a technical feature that empowers users to report inappropriate posts or bad actors to reduce online harm. The deceptively simple flagging interfaces on nearly all major social media platforms disguise…

Large language models (LLMs) have become ubiquitous, thus it is important to understand their risks and limitations. Smaller LLMs can be deployed where compute resources are constrained, such as edge devices, but with different propensity…

We investigate how disagreement in natural language inference (NLI) annotation arises. We developed a taxonomy of disagreement sources with 10 categories spanning 3 high-level classes. We found that some disagreements are due to uncertainty…

Recent applications of LLMs in Machine Reading Comprehension (MRC) systems have shown impressive results, but the use of shortcuts, mechanisms triggered by features spuriously correlated to the true label, has emerged as a potential threat…

As AI agents become more widely deployed, we are likely to see an increasing number of incidents: events involving AI agent use that directly or indirectly cause harm. For example, agents could be prompt-injected to exfiltrate private…

Text classification is a well-studied and versatile building block for many NLP applications. Yet, existing approaches require either large annotated corpora to train a model with or, when using large language models as a base, require…

Dynamic learning systems subject to selective labeling exhibit censoring, i.e. persistent negative predictions assigned to one or more subgroups of points. In applications like consumer finance, this results in groups of applicants that are…

Annotator disagreement is widespread in NLP, particularly for subjective and ambiguous tasks such as toxicity detection and stance analysis. While early approaches treated disagreement as noise to be removed, recent work increasingly models…

Understanding how high-level concepts are represented within artificial neural networks is a fundamental challenge in the field of artificial intelligence. While existing literature in explainable AI emphasizes the importance of labeling…

Many state-of-the-art delineation methods rely on supervised machine learning algorithms. As a result, they require manually annotated training data, which is tedious to obtain. Furthermore, even minor classification errors may…

Backdoor attacks significantly compromise the security of large language models by triggering them to output specific and controlled content. Currently, triggers for textual backdoor attacks fall into two categories: fixed-token triggers…

This paper demonstrates a fatal vulnerability in natural language inference (NLI) and text classification systems. More concretely, we present a 'backdoor poisoning' attack on NLP models. Our poisoning attack utilizes conditional…