In modern society, service robots are increasingly recognized for their wide range of practical applications. In large and crowded social spaces, such as museums and hospitals, these robots are required to safely move in the environment while exhibiting user-friendly behavior. Ensuring the safe and socially acceptable operation of robots in such settings presents several challenges. To enhance the social acceptance in the design process of service robots, we present a systematic analysis of requirements, categorized into functional and non-functional. These requirements are further classified into different categories, with a single requirement potentially belonging to multiple categories. Finally, considering the specific case of a receptionist robotic agent, we discuss the requirements it should possess to ensure social acceptance.

Rapid advances in the capabilities of large language models (LLMs) have raised widespread concerns regarding their potential for malicious use. Open-weight LLMs present unique challenges, as existing safeguards lack robustness to tampering attacks that modify model weights. For example, recent works have demonstrated that refusal and unlearning safeguards can be trivially removed with a few steps of fine-tuning. These vulnerabilities necessitate new approaches for enabling the safe release of open-weight LLMs. We develop a method, called TAR, for building tamper-resistant safeguards into open-weight LLMs such that adversaries cannot remove the safeguards even after thousands of steps of fine-tuning. In extensive evaluations and red teaming analyses, we find that our method greatly improves tamper-resistance while preserving benign capabilities. Our results demonstrate that tamper-resistance is a tractable problem, opening up a promising new avenue to improve the safety and security of open-weight LLMs.

The White House released a statement today outlining commitments that several AI companies are making to curb the creation and distribution of image-based sexual abuse. The participating businesses have laid out the steps they are taking to prevent their platforms from being used to generate non-consensual intimate images (NCII) of adults and child sexual abuse material (CSAM). Specifically, Adobe, Anthropic, Cohere, Common Crawl, Microsoft and OpenAI said they'll be: All of the aforementioned except Common Crawl also agreed they'd be: "incorporating feedback loops and iterative stress-testing strategies in their development processes, to guard against AI models outputting image-based sexual abuse" It's a voluntary commitment, so today's announcement doesn't create any new actionable steps or consequences for failing to follow through on those promises. But it's still worth applauding a good faith effort to tackle this serious problem. The notable absences from today's White House release are Apple, Amazon, Google and Meta. Many big tech and AI companies have been making strides to make it easier for victims of NCII to stop the spread of deepfake images and videos separately from this federal effort.

Online Grooming (OG) is a prevalent threat facing predominately children online, with groomers using deceptive methods to prey on the vulnerability of children on social media/messaging platforms. These attacks can have severe psychological and physical impacts, including a tendency towards revictimization. Current technical measures are inadequate, especially with the advent of end-to-end encryption which hampers message monitoring. Existing solutions focus on the signature analysis of child abuse media, which does not effectively address real-time OG detection. This paper proposes that OG attacks are complex, requiring the identification of specific communication patterns between adults and children. It introduces a novel approach leveraging advanced models such as BERT and RoBERTa for Message-Level Analysis and a Context Determination approach for classifying actor interactions, including the introduction of Actor Significance Thresholds and Message Significance Thresholds. The proposed method aims to enhance accuracy and robustness in detecting OG by considering the dynamic and multi-faceted nature of these attacks. Cross-dataset experiments evaluate the robustness and versatility of our approach. This paper's contributions include improved detection methodologies and the potential for application in various scenarios, addressing gaps in current literature and practices.

Generative AI (GenAI) has been shown to enhance individual productivity in a guided setting. While it is also likely to transform processes in a collaborative work setting, it is unclear what trajectory this transformation will follow. Collaborative environment is characterized by a blend of origination tasks that involve building something from scratch and iteration tasks that involve refining on others' work. Whether GenAI affects these two aspects of collaborative work and to what extent is an open empirical question. We study this question within the open-source development landscape, a prime example of collaborative innovation, where contributions are voluntary and unguided. Specifically, we focus on the launch of GitHub Copilot in October 2021 and leverage a natural experiment in which GitHub Copilot (a programming-focused LLM) selectively rolled out support for Python, but not for R. We observe a significant jump in overall contributions, suggesting that GenAI effectively augments collaborative innovation in an unguided setting. Interestingly, Copilot's launch increased maintenance-related contributions, which are mostly iterative tasks involving building on others' work, significantly more than code-development contributions, which are mostly origination tasks involving standalone contributions. This disparity was exacerbated in active projects with extensive coding activity, raising concerns that, as GenAI models improve to accommodate richer context, the gap between origination and iterative solutions may widen. We discuss practical and policy implications to incentivize high-value innovative solutions.

Large language models (LLMs) have shown considerable success in a range of domain-specific tasks, especially after fine-tuning. However, fine-tuning with real-world data usually leads to privacy risks, particularly when the fine-tuning samples exist in the pre-training data. To avoid the shortcomings of real data, developers often employ methods to automatically generate synthetic data for fine-tuning, as data generated by traditional models are often far away from the real-world pertaining data. However, given the advanced capabilities of LLMs, the distinction between real data and LLM-generated data has become negligible, which may also lead to privacy risks like real data. In this paper, we present an empirical analysis of this underexplored issue by investigating a key question: "Does fine-tuning with LLM-generated data enhance privacy, or does it pose additional privacy risks?" Based on the structure of LLM's generated data, our research focuses on two primary approaches to fine-tuning with generated data: supervised fine-tuning with unstructured generated data and self-instruct tuning. The number of successful Personal Information Identifier (PII) extractions for Pythia after fine-tuning our generated data raised over $20\%$. Furthermore, the ROC-AUC score of membership inference attacks for Pythia-6.9b after self-instruct methods also achieves more than $40\%$ improvements on ROC-AUC score than base models. The results indicate the potential privacy risks in LLMs when fine-tuning with the generated data.

The focus of this study is to evaluate the effectiveness of Machine Learning (ML) methods for two-sample testing with right-censored observations. To achieve this, we develop several ML-based methods with varying architectures and implement them as two-sample tests. Each method is an ensemble (stacking) that combines predictions from classical two-sample tests. This paper presents the results of training the proposed ML methods, examines their statistical power compared to classical two-sample tests, analyzes the distribution of test statistics for the proposed methods when the null hypothesis is true, and evaluates the significance of the features incorporated into the proposed methods. All results from numerical experiments were obtained from a synthetic dataset generated using the Smirnov transform (Inverse Transform Sampling) and replicated multiple times through Monte Carlo simulation. To test the two-sample problem with right-censored observations, one can use the proposed two-sample methods. All necessary materials (source code, example scripts, dataset, and samples) are available on GitHub and Hugging Face.

AI Safety is an emerging area of critical importance to the safe adoption and deployment of AI systems. With the rapid proliferation of AI and especially with the recent advancement of Generative AI (or GAI), the technology ecosystem behind the design, development, adoption, and deployment of AI systems has drastically changed, broadening the scope of AI Safety to address impacts on public safety and national security. In this paper, we propose a novel architectural framework for understanding and analyzing AI Safety; defining its characteristics from three perspectives: Trustworthy AI, Responsible AI, and Safe AI. We provide an extensive review of current research and advancements in AI safety from these perspectives, highlighting their key challenges and mitigation approaches. Through examples from state-of-the-art technologies, particularly Large Language Models (LLMs), we present innovative mechanism, methodologies, and techniques for designing and testing AI safety. Our goal is to promote advancement in AI safety research, and ultimately enhance people's trust in digital transformation.

This paper explores the intersection of technological innovation and access to justice by developing a benchmark for predicting case outcomes in the UK Employment Tribunal (UKET). To address the challenge of extensive manual annotation, the study employs a large language model (LLM) for automatic annotation, resulting in the creation of the CLC-UKET dataset. The dataset consists of approximately 19,000 UKET cases and their metadata. Comprehensive legal annotations cover facts, claims, precedent references, statutory references, case outcomes, reasons and jurisdiction codes. Facilitated by the CLC-UKET data, we examine a multi-class case outcome prediction task in the UKET. Human predictions are collected to establish a performance reference for model comparison. Empirical results from baseline models indicate that finetuned transformer models outperform zero-shot and few-shot LLMs on the UKET prediction task. The performance of zero-shot LLMs can be enhanced by integrating task-related information into few-shot examples. We hope that the CLC-UKET dataset, along with human annotations and empirical findings, can serve as a valuable benchmark for employment-related dispute resolution.

Nevada has a new helper in its quest to plow through a backlog of unemployment claims: Google AI. Gizmodo reports that the initiative will task one of the company's cloud-based AI models with analyzing appeals hearing transcripts and suggesting whether cases should be approved. Welcome to the future, where a robot weighs in on whether you get the government money you requested. The Nevada Independent wrote in June that the AI model, trained on the state's unemployment law and policies, will analyze transcripts of virtual appeals hearings. It will then spit out a ruling, which a state employee will review for mistakes and decide whether to honor.