The feud between Musk and Altman has, over time, encompassed a slew of lawsuits, email dumps and social media digs. Their rivalry can be traced back a decade, with Musk's now public belief that OpenAI, under Altman's leadership, abandoned the principles he and others used to found it in 2015. The firm was created with the intention of building artificial general intelligence (AGI) - AI that can perform any task that a human being is capable of - but by making its technology open-source and promising to "benefit humanity". OpenAI was also set up as a not-for-profit company, meaning it would not aim to make money, but in 2019 it established a for-profit arm which Musk felt was antithetical to its original mission. Musk argued in his March 2024 lawsuit that the firm had instead been focusing on "maximising profits" for its major investor Microsoft.

Gone are the days when the web was dominated by humans posting social media updates or exchanging memes. Earlier this year, for the first time since the data has been tracked, web-browsing bots, rather than humans, accounted for the bulk of web traffic. Well over half of that bot traffic is from malicious bots, hoovering up personal data left unprotected online, for instance. But an increasing proportion comes from bots sent out by artificial intelligence companies to gather data for their models or respond to user prompts. Indeed, ChatGPT-User, a bot powering OpenAI's ChatGPT, is now responsible for 6 per cent of all web traffic, while ClaudeBot, an automated system developed by AI company Anthropic, accounts for 13 per cent.

HMRC has confirmed it uses artificial intelligence (AI) to monitor social media posts as part of criminal investigations into suspected tax cheats. It said the tech would not replace "human decision-making" and was subject to legal oversight. "Greater use of AI will enable our staff to spend less time on admin and more time helping taxpayers, as well as better target fraud and evasion to bring in more money for public services," it said in a statement. However, experts warn there are risks with using AI in this way.

The propensity for AI systems to make mistakes that humans miss has been on full display in the US legal system as of late. The follies began when lawyers submitted documents citing cases that didn't exist. Similar mistakes soon spread to other roles in the courts. Last December, a Stanford professor submitted sworn testimony containing hallucinations and errors in a case about deepfakes, despite being an expert on AI and misinformation himself. Now, judges are experimenting with generative AI too. Some believe that with the right precautions, the technology can expedite legal research, summarize cases, draft routine orders, and overall help speed up the court system, which is badly backlogged in many parts of the US.

Apple's AI Ambitions Leave Big Questions Over Its Climate Goals Halfway to its 2030 net-zero goal, Apple faces slow and hold-out suppliers, a tariffs scramble, and an AI race that could profoundly impact eco-friendly ambitions. Here's a simple question: Is the current top iPhone better for the environment than the top iPhone was five years ago? Let's take the iPhone Pro series. If we're looking at recycled and renewable materials, it's an easy yes. Compare the iPhone 11 Pro, released in September 2019, with the iPhone 16 Pro, released in September 2024, and there has been good progress--from a few smaller components and packaging to now at more than 25 percent of the whole phone.

Machine learning (ML) systems are increasingly deployed in high-stakes domains where reliability is paramount. This thesis investigates how uncertainty estimation can enhance the safety and trustworthiness of ML, focusing on selective prediction -- where models abstain when confidence is low. We first show that a model's training trajectory contains rich uncertainty signals that can be exploited without altering its architecture or loss. By ensembling predictions from intermediate checkpoints, we propose a lightweight, post-hoc abstention method that works across tasks, avoids the cost of deep ensembles, and achieves state-of-the-art selective prediction performance. Crucially, this approach is fully compatible with differential privacy (DP), allowing us to study how privacy noise affects uncertainty quality. We find that while many methods degrade under DP, our trajectory-based approach remains robust, and we introduce a framework for isolating the privacy-uncertainty trade-off. Next, we then develop a finite-sample decomposition of the selective classification gap -- the deviation from the oracle accuracy-coverage curve -- identifying five interpretable error sources and clarifying which interventions can close the gap. This explains why calibration alone cannot fix ranking errors, motivating methods that improve uncertainty ordering. Finally, we show that uncertainty signals can be adversarially manipulated to hide errors or deny service while maintaining high accuracy, and we design defenses combining calibration audits with verifiable inference. Together, these contributions advance reliable ML by improving, evaluating, and safeguarding uncertainty estimation, enabling models that not only make accurate predictions -- but also know when to say "I do not know".

Deep neural networks (DNNs) are being utilized in various aspects of our daily lives, including high-stakes decision-making applications that impact individuals. However, these systems reflect and amplify bias from the data used during training and testing, potentially resulting in biased behavior and inaccurate decisions. For instance, having different misclassification rates between white and black sub-populations. However, effectively and efficiently identifying and correcting biased behavior in DNNs is a challenge. This paper introduces FairFLRep, an automated fairness-aware fault localization and repair technique that identifies and corrects potentially bias-inducing neurons in DNN classifiers. FairFLRep focuses on adjusting neuron weights associated with sensitive attributes, such as race or gender, that contribute to unfair decisions. By analyzing the input-output relationships within the network, FairFLRep corrects neurons responsible for disparities in predictive quality parity. We evaluate FairFLRep on four image classification datasets using two DNN classifiers, and four tabular datasets with a DNN model. The results show that FairFLRep consistently outperforms existing methods in improving fairness while preserving accuracy. An ablation study confirms the importance of considering fairness during both fault localization and repair stages. Our findings also show that FairFLRep is more efficient than the baseline approaches in repairing the network.

Subjective language understanding refers to a broad set of natural language processing tasks where the goal is to interpret or generate content that conveys personal feelings, opinions, or figurative meanings rather than objective facts. With the advent of large language models (LLMs) such as ChatGPT, LLaMA, and others, there has been a paradigm shift in how we approach these inherently nuanced tasks. In this survey, we provide a comprehensive review of recent advances in applying LLMs to subjective language tasks, including sentiment analysis, emotion recognition, sarcasm detection, humor understanding, stance detection, metaphor interpretation, intent detection, and aesthetics assessment. We begin by clarifying the definition of subjective language from linguistic and cognitive perspectives, and we outline the unique challenges posed by subjective language (e.g. ambiguity, figurativeness, context dependence). We then survey the evolution of LLM architectures and techniques that particularly benefit subjectivity tasks, highlighting why LLMs are well-suited to model subtle human-like judgments. For each of the eight tasks, we summarize task definitions, key datasets, state-of-the-art LLM-based methods, and remaining challenges. We provide comparative insights, discussing commonalities and differences among tasks and how multi-task LLM approaches might yield unified models of subjectivity. Finally, we identify open issues such as data limitations, model bias, and ethical considerations, and suggest future research directions. We hope this survey will serve as a valuable resource for researchers and practitioners interested in the intersection of affective computing, figurative language processing, and large-scale language models.

Forensic cause-of-death determination faces systemic challenges, including workforce shortages and diagnostic variability, particularly in high-volume systems like China's medicolegal infrastructure. We introduce FEAT (ForEnsic AgenT), a multi-agent AI framework that automates and standardizes death investigations through a domain-adapted large language model. FEAT's application-oriented architecture integrates: (i) a central Planner for task decomposition, (ii) specialized Local Solvers for evidence analysis, (iii) a Memory & Reflection module for iterative refinement, and (iv) a Global Solver for conclusion synthesis. The system employs tool-augmented reasoning, hierarchical retrieval-augmented generation, forensic-tuned LLMs, and human-in-the-loop feedback to ensure legal and medical validity. In evaluations across diverse Chinese case cohorts, FEAT outperformed state-of-the-art AI systems in both long-form autopsy analyses and concise cause-of-death conclusions. It demonstrated robust generalization across six geographic regions and achieved high expert concordance in blinded validations. Senior pathologists validated FEAT's outputs as comparable to those of human experts, with improved detection of subtle evidentiary nuances. To our knowledge, FEAT is the first LLM-based AI agent system dedicated to forensic medicine, offering scalable, consistent death certification while maintaining expert-level rigor. By integrating AI efficiency with human oversight, this work could advance equitable access to reliable medicolegal services while addressing critical capacity constraints in forensic systems.

Federated unlearning (FU) algorithms allow clients in federated settings to exercise their ''right to be forgotten'' by removing the influence of their data from a collaboratively trained model. Existing FU methods maintain data privacy by performing unlearning locally on the client-side and sending targeted updates to the server without exposing forgotten data; yet they often rely on server-side cooperation, revealing the client's intent and identity without enforcement guarantees - compromising autonomy and unlearning privacy. In this work, we propose EFU (Enforced Federated Unlearning), a cryptographically enforced FU framework that enables clients to initiate unlearning while concealing its occurrence from the server. Specifically, EFU leverages functional encryption to bind encrypted updates to specific aggregation functions, ensuring the server can neither perform unauthorized computations nor detect or skip unlearning requests. To further mask behavioral and parameter shifts in the aggregated model, we incorporate auxiliary unlearning losses based on adversarial examples and parameter importance regularization. Extensive experiments show that EFU achieves near-random accuracy on forgotten data while maintaining performance comparable to full retraining across datasets and neural architectures - all while concealing unlearning intent from the server. Furthermore, we demonstrate that EFU is agnostic to the underlying unlearning algorithm, enabling secure, function-hiding, and verifiable unlearning for any client-side FU mechanism that issues targeted updates.