Federated learning facilitates collaborative machine learning, enabling multiple participants to collectively develop a shared model while preserving the privacy of individual data. The growing importance of the "right to be forgotten" calls for effective mechanisms to facilitate data removal upon request. In response, federated unlearning (FU) has been developed to efficiently eliminate the influence of specific data from the model. Current FU methods primarily rely on approximate unlearning strategies, which seek to balance data removal efficacy with computational and communication costs, but often fail to completely erase data influence. To address these limitations, we propose FedUHB, a novel exact unlearning approach that leverages the Polyak heavy ball optimization technique, a first-order method, to achieve rapid retraining. In addition, we introduce a dynamic stopping mechanism to optimize the termination of the unlearning process. Our extensive experiments show that FedUHB not only enhances unlearning efficiency but also preserves robust model performance after unlearning. Furthermore, the dynamic stopping mechanism effectively reduces the number of unlearning iterations, conserving both computational and communication resources. FedUHB can be proved as an effective and efficient solution for exact data removal in federated learning settings.

Federated unlearning (FU) offers a promising solution to effectively address the need to erase the impact of specific clients' data on the global model in federated learning (FL), thereby granting individuals the ``Right to be Forgotten". The most straightforward approach to achieve unlearning is to train the model from scratch, excluding clients who request data removal, but it is resource-intensive. Current state-of-the-art FU methods extend traditional FL frameworks by leveraging stored historical updates, enabling more efficient unlearning than training from scratch. However, the use of stored updates introduces significant privacy risks. Adversaries with access to these updates can potentially reconstruct clients' local data, a well-known vulnerability in the privacy domain. While privacy-enhanced techniques exist, their applications to FU scenarios that balance unlearning efficiency with privacy protection remain underexplored. To address this gap, we propose FedADP, a method designed to achieve both efficiency and privacy preservation in FU. Our approach incorporates an adaptive differential privacy (DP) mechanism, carefully balancing privacy and unlearning performance through a novel budget allocation strategy tailored for FU. FedADP also employs a dual-layered selection process, focusing on global models with significant changes and client updates closely aligned with the global model, reducing storage and communication costs. Additionally, a novel calibration method is introduced to facilitate effective unlearning. Extensive experimental results demonstrate that FedADP effectively manages the trade-off between unlearning efficiency and privacy protection.

Sparse Autoencoders for transformer-based language models are typically defined independently per layer. In this work we analyze statistical relationships between features in adjacent layers to understand how features evolve through a forward pass. We provide a graph visualization interface for features and their most similar next-layer neighbors, and build communities of related features across layers. We find that a considerable amount of features are passed through from a previous layer, some features can be expressed as quasi-boolean combinations of previous features, and some features become more specialized in later layers.

In perhaps the most adorable hacker story of the year, a trio of technologists in India found an innovative way to circumvent Apple's location restrictions on AirPod Pro 2s so they could enable the earbuds' hearing aid feature for their grandmas. The hack involved a homemade Faraday cage, a microwave, and a lot of trial and error. On the other end of the tech-advancements spectrum, the US military is currently testing an AI-enabled machine gun that is capable of auto-targeting swarms of drones. The Bullfrog, built by Allen Control Systems, is one of several advanced weapons technologies in the works to combat the growing threat of cheap, small drones on the battlefield. The US Department of Justice announced this week that an 18-year-old from California has admitted to making or orchestrating more than 375 swatting attacks across the United States.

This paper presents a framework for developing efficient and interpretable carbon-cycle emulators (CCEs) as part of climate emulators in Integrated Assessment Models, enabling economists to custom-build CCEs accurately calibrated to advanced climate science. We propose a generalized multi-reservoir linear box-model CCE that preserves key physical quantities and can be use-case tailored for specific use cases. Three CCEs are presented for illustration: the 3SR model (replicating DICE-2016), the 4PR model (including the land biosphere), and the 4PR-X model (accounting for dynamic land-use changes like deforestation that impact the reservoir's storage capacity). Evaluation of these models within the DICE framework shows that land-use changes in the 4PR-X model significantly impact atmospheric carbon and temperatures -- emphasizing the importance of using tailored climate emulators. By providing a transparent and flexible tool for policy analysis, our framework allows economists to assess the economic impacts of climate policies more accurately.

Due to increasing privacy regulations and regulatory compliance, Machine Unlearning (MU) has become essential. The goal of unlearning is to remove information related to a specific class from a model. Traditional approaches achieve exact unlearning by retraining the model on the remaining dataset, but incur high computational costs. This has driven the development of more efficient unlearning techniques, including model sparsification techniques, which boost computational efficiency, but degrade the model's performance on the remaining classes. To mitigate these issues, we propose a novel method, PruneLoRA which introduces a new MU paradigm, termed prune first, then adapt, then unlearn. LoRA (Hu et al. 2022) reduces the need for large-scale parameter updates by applying low-rank updates to the model. We leverage LoRA to selectively modify a subset of the pruned model's parameters, thereby reducing the computational cost, memory requirements and improving the model's ability to retain performance on the remaining classes. Experimental Results across various metrics showcase that our method outperforms other approximate MU methods and bridges the gap between exact and approximate unlearning. Our code is available at https://github.com/vlgiitr/LoRA-Unlearn.

Large Language Models (LLMs) have revolutionized natural language processing, but their susceptibility to biases poses significant challenges. This comprehensive review examines the landscape of bias in LLMs, from its origins to current mitigation strategies. We categorize biases as intrinsic and extrinsic, analyzing their manifestations in various NLP tasks. The review critically assesses a range of bias evaluation methods, including data-level, model-level, and output-level approaches, providing researchers with a robust toolkit for bias detection. We further explore mitigation strategies, categorizing them into pre-model, intra-model, and post-model techniques, highlighting their effectiveness and limitations. Ethical and legal implications of biased LLMs are discussed, emphasizing potential harms in real-world applications such as healthcare and criminal justice. By synthesizing current knowledge on bias in LLMs, this review contributes to the ongoing effort to develop fair and responsible AI systems. Our work serves as a comprehensive resource for researchers and practitioners working towards understanding, evaluating, and mitigating bias in LLMs, fostering the development of more equitable AI technologies.

Several GenAI coding assistants, including GitHub's Copilot [45], Tabnine [119], Codeium [24], and Cody [25], as well as general purpose tools such as ChatGPT [100], Claude [11], and Gemini [42], have become readily accessible, either as IDE extensions or standalone applications, enabling developers to perform many coding tasks with little effort, including automated code completion, summarization, and debugging.

Elon Musk has expanded his lawsuit against the ChatGPT maker OpenAI, adding federal antitrust and other claims and adding OpenAI's largest financial backer, Microsoft, as a defendant. Musk's amended lawsuit, filed on Thursday night in federal court in Oakland, California, said Microsoft and OpenAI illegally sought to monopolize the market for generative artificial intelligence and sideline competitors. Like Musk's original August complaint, it accused OpenAI and its chief executive, Samuel Altman, of violating contract provisions by putting profits ahead of the public good in the push to advance AI. "Never before has a corporation gone from tax-exempt charity to a 157bn for-profit, market-paralyzing gorgon – and in just eight years," the complaint said. It seeks to void OpenAI's license with Microsoft and force them to divest "ill-gotten" gains. OpenAI in a statement said the latest lawsuit "is even more baseless and overreaching than the previous ones".

Elon Musk's X is taking the state of California to court over a new law that prevents the spread of AI-generated election misinformation. Bloomberg reports that X filed a lawsuit against AB 2655, also known as the Defending Democracy from Deepfake Deception Act of 2024, in a Sacramento federal court. California Gov. Gavin Newsom signed the bill into law on September 17, creating accountability standards for using false political speech faked with AI programs close to an election. The legislation prevents the distribution of "materially deceptive audio or visual media of a candidate within 60 days of an election at which the candidate will appear on the ballet." X argues that the law will create more political speech censorship.