arXiv subjects

Rajvardhan Oak

Publications and source records attributed to Rajvardhan Oak.

6 recordsLinked to original sources

arXiv2026

Muse Spark Safety & Preparedness Report

Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refusal across a range of benchmarks related to hazardous workflows in chemistry and biology. We therefore release Muse Spark as the underlying model of Meta AI.

cs.CY↗

arXiv2025

Re-ranking Using Large Language Models for Mitigating Exposure to Harmful Content on Social Media Platforms

Social media platforms utilize Machine Learning (ML) and Artificial Intelligence (AI) powered recommendation algorithms to maximize user engagement, which can result in inadvertent exposure to harmful content. Current moderation efforts, reliant on classifiers trained with extensive human-annotated data, struggle with scalability and adapting to new forms of harm. To address these challenges, we propose a novel re-ranking approach using Large Language Models (LLMs) in zero-shot and few-shot settings. Our method dynamically assesses and re-ranks content sequences, effectively mitigating harmful content exposure without requiring extensive labeled data. Alongside traditional ranking metrics, we also introduce two new metrics to evaluate the effectiveness of re-ranking in reducing exposure to harmful content. Through experiments on three datasets, three models and across three configurations, we demonstrate that our LLM-based approach significantly outperforms existing proprietary moderation approaches, offering a scalable and adaptable solution for harm mitigation.

cs.CL↗

arXiv2025

"Hello, is this Anna?": Unpacking the Lifecycle of Pig-Butchering Scams

Pig-butchering scams have emerged as a complex form of fraud that combines elements of romance, investment fraud, and advanced social engineering tactics to systematically exploit victims. In this paper, we present the first qualitative analysis of pig-butchering scams, informed by in-depth semi-structured interviews with $N=26$ victims. We capture nuanced, first-hand accounts from victims, providing insight into the lifecycle of pig-butchering scams and the complex emotional and financial manipulation involved. We systematically analyze each phase of the scam, revealing that perpetrators employ tactics such as staged trust-building, fraudulent financial platforms, fabricated investment returns, and repeated high-pressure tactics, all designed to exploit victims' trust and financial resources over extended periods. Our findings reveal an organized scam lifecycle characterized by emotional manipulation, staged financial exploitation, and persistent re-engagement efforts that amplify victim losses. We also find complex psychological and financial impacts on victims, including heightened vulnerability to secondary scams. Finally, we propose actionable intervention points for social media and financial platforms to curb the prevalence of these scams and highlight the need for non-stigmatizing terminology to encourage victims to report and seek assistance.

cs.CR↗

arXiv2025

Towards Safer Social Media Platforms: Scalable and Performant Few-Shot Harmful Content Moderation Using Large Language Models

The prevalence of harmful content on social media platforms poses significant risks to users and society, necessitating more effective and scalable content moderation strategies. Current approaches rely on human moderators, supervised classifiers, and large volumes of training data, and often struggle with scalability, subjectivity, and the dynamic nature of harmful content (e.g., violent content, dangerous challenge trends, etc.). To bridge these gaps, we utilize Large Language Models (LLMs) to undertake few-shot dynamic content moderation via in-context learning. Through extensive experiments on multiple LLMs, we demonstrate that our few-shot approaches can outperform existing proprietary baselines (Perspective and OpenAI Moderation) as well as prior state-of-the-art few-shot learning methods, in identifying harm. We also incorporate visual information (video thumbnails) and assess if different multimodal techniques improve model performance. Our results underscore the significant benefits of employing LLM based methods for scalable and dynamic harmful content moderation online.

cs.CL↗

arXiv2024

Understanding Underground Incentivized Review Services

While human factors in fraud have been studied by the HCI and security communities, most research has been directed to understanding either the victims' perspectives or prevention strategies, and not on fraudsters, their motivations and operation techniques. Additionally, the focus has been on a narrow set of problems: phishing, spam and bullying. In this work, we seek to understand review fraud on e-commerce platforms through an HCI lens. Through surveys with real fraudsters (N=36 agents and N=38 reviewers), we uncover sophisticated recruitment, execution, and reporting mechanisms fraudsters use to scale their operation while resisting takedown attempts, including the use of AI tools like ChatGPT. We find that countermeasures that crack down on communication channels through which these services operate are effective in combating incentivized reviews. This research sheds light on the complex landscape of incentivized reviews, providing insights into the mechanics of underground services and their resilience to removal efforts.

cs.CY↗

arXiv2020

Time for a Background Check! Uncovering the impact of Background Features on Deep Neural Networks

With increasing expressive power, deep neural networks have significantly improved the state-of-the-art on image classification datasets, such as ImageNet. In this paper, we investigate to what extent the increasing performance of deep neural networks is impacted by background features? In particular, we focus on background invariance, i.e., accuracy unaffected by switching background features and background influence, i.e., predictive power of background features itself when foreground is masked. We perform experiments with 32 different neural networks ranging from small-size networks to large-scale networks trained with up to one Billion images. Our investigations reveal that increasing expressive power of DNNs leads to higher influence of background features, while simultaneously, increases their ability to make the correct prediction when background features are removed or replaced with a randomly selected texture-based background.

cs.CV↗