Large language models (LLMs) have the potential to transform the practice of law, but this potential is threatened by the presence of legal hallucinations -- responses from these models that are not consistent with legal facts. We investigate the extent of these hallucinations using an original suite of legal queries, comparing LLMs' responses to structured legal metadata and examining their consistency. Our work makes four key contributions: (1) We develop a typology of legal hallucinations, providing a conceptual framework for future research in this area. (2) We find that legal hallucinations are alarmingly prevalent, occurring between 69% of the time with ChatGPT 3.5 and 88% with Llama 2, when these models are asked specific, verifiable questions about random federal court cases. (3) We illustrate that LLMs often fail to correct a user's incorrect legal assumptions in a contra-factual question setup. (4) We provide evidence that LLMs cannot always predict, or do not always know, when they are producing legal hallucinations. Taken together, these findings caution against the rapid and unsupervised integration of popular LLMs into legal tasks. Even experienced lawyers must remain wary of legal hallucinations, and the risks are highest for those who stand to benefit from LLMs the most -- pro se litigants or those without access to traditional legal resources.

We propose and evaluate an automated pipeline for discovering significant topics from legal decision texts by passing features synthesized with topic models through penalised regressions and post-selection significance tests. The method identifies case topics significantly correlated with outcomes, topic-word distributions which can be manually-interpreted to gain insights about significant topics, and case-topic weights which can be used to identify representative cases for each topic. We demonstrate the method on a new dataset of domain name disputes and a canonical dataset of European Court of Human Rights violation cases. Topic models based on latent semantic analysis as well as language model embeddings are evaluated. We show that topics derived by the pipeline are consistent with legal doctrines in both areas and can be useful in other related legal analysis tasks.

AI alignment aims to make AI systems behave in line with human intentions and values. As AI systems grow more capable, so do risks from misalignment. To provide a comprehensive and up-to-date overview of the alignment field, in this survey, we delve into the core concepts, methodology, and practice of alignment. First, we identify four principles as the key objectives of AI alignment: Robustness, Interpretability, Controllability, and Ethicality (RICE). Guided by these four principles, we outline the landscape of current alignment research and decompose them into two key components: forward alignment and backward alignment. The former aims to make AI systems aligned via alignment training, while the latter aims to gain evidence about the systems' alignment and govern them appropriately to avoid exacerbating misalignment risks. On forward alignment, we discuss techniques for learning from feedback and learning under distribution shift. On backward alignment, we discuss assurance techniques and governance practices. We also release and continually update the website (www.alignmentsurvey.com) which features tutorials, collections of papers, blog posts, and other resources.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. With a wary eye over the future of the federal courts, Supreme Court Chief Justice John Roberts warned Sunday of the perils of artificial intelligence (AI) when deciding cases and other important legal matters. His remarks came in the annual year-end report issued by the head of the federal judiciary, which made no mention of current controversies surrounding his court, including calls for greater transparency and ethics reform binding the justices. Noting the legal profession in general is "notoriously averse to change," Roberts urged a go-slow approach when embracing new technologies by the courts.

Two women have created an AI chat bot to allow individuals in the workplace to easily find out if they are victims of sexual harassment. The pioneering tool, which is aimed at helping victims anonymously report both discrimination and racism as well as sexual harassment, allows individuals to ask personally-curated questions for an AI bot to assess and answer. Trained on the UK Equality Act, workers can ask questions like: 'My boss keeps asking me to have dinner with him and stroking my arm. I have said no several times and it's making me anxious. The tool is part of an app called'SaferSpace', founded by PR guru Ruth Sparkes and business entrepreneur Sunita Gordon.

Jefferson County Sheriff Jeff Bullard said after one year in effect, the SAFE-T Act is having the "intended result" and damaging the policing profession in Illinois. With the calendar-page turn to 2024 on Monday comes 320 new state laws that Illinois residents will need to navigate. Some will have a widespread effect, including a law banning semi-automatic rifles and another requiring paid time off. But others won't have an immediate or noticeable impact, including a law that lets county governments consider a potential contractor's participation in an approved apprenticeship program in determining the winning low bid for a project. One law that took effect in 2019 but is still impacting tens of thousands of workers is an increase in the minimum wage.

We introduce the Cambridge Law Corpus (CLC), a dataset for legal AI research. It consists of over 250 000 court cases from the UK. Most cases are from the 21st century, but the corpus includes cases as old as the 16th century. This paper presents the first release of the corpus, containing the raw text and meta-data. Together with the corpus, we provide annotations on case outcomes for 638 cases, done by legal experts. Using our annotated data, we have trained and evaluated case outcome extraction with GPT-3, GPT-4 and RoBERTa models to provide benchmarks. We include an extensive legal and ethical discussion to address the potentially sensitive nature of this material. As a consequence, the corpus will only be released for research purposes under certain restrictions.

The revolution of natural language processing via large language models has motivated its use in multidisciplinary areas that include social sciences and humanities and more specifically, comparative religion. Sentiment analysis provides a mechanism to study the emotions expressed in text. Recently, sentiment analysis has been used to study and compare translations of the Bhagavad Gita, which is a fundamental and sacred Hindu text. In this study, we use sentiment analysis for studying selected chapters of the Bible. These chapters are known as the Sermon on the Mount. We utilize a pre-trained language model for sentiment analysis by reviewing five translations of the Sermon on the Mount, which include the King James version, the New International Version, the New Revised Standard Version, the Lamsa Version, and the Basic English Version. We provide a chapter-by-chapter and verse-by-verse comparison using sentiment and semantic analysis and review the major sentiments expressed. Our results highlight the varying sentiments across the chapters and verses. We found that the vocabulary of the respective translations is significantly different. We detected different levels of humour, optimism, and empathy in the respective chapters that were used by Jesus to deliver his message.

Pre-training, which utilizes extensive and varied datasets, is a critical factor in the success of Large Language Models (LLMs) across numerous applications. However, the detailed makeup of these datasets is often not disclosed, leading to concerns about data security and potential misuse. This is particularly relevant when copyrighted material, still under legal protection, is used inappropriately, either intentionally or unintentionally, infringing on the rights of the authors. In this paper, we introduce a detailed framework designed to detect and assess the presence of content from potentially copyrighted books within the training datasets of LLMs. This framework also provides a confidence estimation for the likelihood of each content sample's inclusion. To validate our approach, we conduct a series of simulated experiments, the results of which affirm the framework's effectiveness in identifying and addressing instances of content misuse in LLM training processes. Furthermore, we investigate the presence of recognizable quotes from famous literary works within these datasets. The outcomes of our study have significant implications for ensuring the ethical use of copyrighted materials in the development of LLMs, highlighting the need for more transparent and responsible data management practices in this field.

With the development of deep learning and natural language processing techniques, pre-trained language models have been widely used to solve information retrieval (IR) problems. Benefiting from the pre-training and fine-tuning paradigm, these models achieve state-of-the-art performance. In previous works, plain texts in Wikipedia have been widely used in the pre-training stage. However, the rich structured information in Wikipedia, such as the titles, abstracts, hierarchical heading (multi-level title) structure, relationship between articles, references, hyperlink structures, and the writing organizations, has not been fully explored. In this paper, we devise four pre-training objectives tailored for IR tasks based on the structured knowledge of Wikipedia. Compared to existing pre-training methods, our approach can better capture the semantic knowledge in the training corpus by leveraging the human-edited structured data from Wikipedia. Experimental results on multiple IR benchmark datasets show the superior performance of our model in both zero-shot and fine-tuning settings compared to existing strong retrieval baselines. Besides, experimental results in biomedical and legal domains demonstrate that our approach achieves better performance in vertical domains compared to previous models, especially in scenarios where long text similarity matching is needed.