Maybe it was the sight of Sengled users literally left in the dark by their useless Wi-Fi bulbs, maybe it was another price hike, or just an overall sense that my smart devices weren't truly under my control. Whatever the reason, I'd developed a growing desire to build a smart home setup that wasn't a hostage to the cloud. Specifically, I'm talking about a locally hosted smart home setup, and I'm currently in the process of building one. And while I'm a smart home expert thanks to my six years' experience here at TechHive, I'm quickly realizing how much I still don't know as I tackle the steep learning curve of a DIY smart home. This isn't a step-by-step guide of how to build your own smart home system--that might come later--but more of a journal about where I am in my self-hosted smart home journey, where I started, and what I'm hoping to achieve.

A new report by the Wall Street Journal is warning people to think twice before trusting the'unsubscribe' option at the bottom of promotional emails. Scammers are working harder than ever to trick people online, but WhatsApp is hitting back in a big way. In the first half of 2025, WhatsApp removed more than 6.8 million accounts tied to global scam networks, many of them run by criminal groups and even forced labor camps in Southeast Asia. At the same time, the app is rolling out new tools to make it easier to spot scams and stay safe while messaging. These updates come as fraudsters increasingly use private messaging apps, dating platforms, social media and even AI tools like ChatGPT to run their schemes.

Physicists have created a new type of radar that could help improve underground imaging, using a cloud of atoms in a glass cell to detect reflected radio waves. The radar is a type of quantum sensor, an emerging technology that uses the quantum-mechanical properties of objects as measurement devices. It's still a prototype, but its intended use is to image buried objects in situations such as constructing underground utilities, drilling wells for natural gas, and excavating archaeological sites. If you're interested in the potential of quantum, why not check out: Why AI could eat quantum computing's lunch. Rapid advances in applying artificial intelligence to simulations in physics and chemistry have some people questioning whether we will even need quantum computers at all.

Artificial intelligence tools used by more than half of England's councils are downplaying women's physical and mental health issues and risk creating gender bias in care decisions, research has found. The study found that when using Google's AI tool "Gemma" to generate and summarise the same case notes, language such as "disabled", "unable" and "complex" appeared significantly more often in descriptions of men than women. The study, by the London School of Economics and Political Science (LSE), also found that similar care needs in women were more likely to be omitted or described in less serious terms. Dr Sam Rickman, the lead author of the report and a researcher in LSE's Care Policy and Evaluation Centre, said AI could result in "unequal care provision for women". "We know these models are being used very widely and what's concerning is that we found very meaningful differences between measures of bias in different models," he said.

Reinforcement learning (RL)-based fine-tuning has become a crucial step in post-training language models for advanced mathematical reasoning and coding. Following the success of frontier reasoning models, recent work has demonstrated that RL fine-tuning consistently improves performance, even in smaller-scale models; however, the underlying mechanisms driving these improvements are not well-understood. Understanding the effects of RL fine-tuning requires disentangling its interaction with pretraining data composition, hyperparameters, and model scale, but such problems are exacerbated by the lack of transparency regarding the training data used in many existing models. In this work, we present a systematic end-to-end study of RL fine-tuning for mathematical reasoning by training models entirely from scratch on different mixtures of fully open datasets. We investigate the effects of various RL fine-tuning algorithms (PPO, GRPO, and Expert Iteration) across models of different scales. Our study reveals that RL algorithms consistently converge towards a dominant output distribution, amplifying patterns in the pretraining data. We also find that models of different scales trained on the same data mixture will converge to distinct output distributions, suggesting that there are scale-dependent biases in model generalization. Moreover, we find that RL post-training on simpler questions can lead to performance gains on harder ones, indicating that certain reasoning capabilities generalize across tasks. Our findings show that small-scale proxies in controlled settings can elicit interesting insights regarding the role of RL in shaping language model behavior.

The preservation of privacy has emerged as a critical topic in the era of artificial intelligence. However, current work focuses on user-oriented privacy, overlooking severe enterprise data leakage risks exacerbated by the Retrieval-Augmented Generation paradigm. To address this gap, our paper introduces a novel objective: enterprise-oriented privacy concerns. Achieving this objective requires overcoming two fundamental challenges: existing methods such as data sanitization severely degrade model performance, and the field lacks public datasets for evaluation. We address these challenges with several solutions. (1) To prevent performance degradation, we propose ABack, a training-free mechanism that leverages a Hidden State Model to pinpoint the origin of a leakage intention and rewrite the output safely. (2) To solve the lack of datasets, we construct PriGenQA, a new benchmark for enterprise privacy scenarios in healthcare and finance. To ensure a rigorous evaluation, we move beyond simple static attacks by developing a powerful adaptive attacker with Group Relative Policy Optimization. Experiments show that against this superior adversary, ABack improves the overall privacy utility score by up to 15\% over strong baselines, avoiding the performance trade-offs of prior methods.

As large language models (LLMs) continue to advance and gain influence, the development of embodied AI has accelerated, drawing significant attention, particularly in navigation scenarios. Embodied navigation requires an agent to perceive, interact with, and adapt to its environment while moving toward a specified target in unfamiliar settings. However, the integration of embodied navigation into critical applications raises substantial safety concerns. Given their deployment in dynamic, real-world environments, ensuring the safety of such systems is critical. This survey provides a comprehensive analysis of safety in embodied navigation from multiple perspectives, encompassing attack strategies, defense mechanisms, and evaluation methodologies. Beyond conducting a comprehensive examination of existing safety challenges, mitigation technologies, and various datasets and metrics that assess effectiveness and robustness, we explore unresolved issues and future research directions in embodied navigation safety. These include potential attack methods, mitigation strategies, more reliable evaluation techniques, and the implementation of verification frameworks. By addressing these critical gaps, this survey aims to provide valuable insights that can guide future research toward the development of safer and more reliable embodied navigation systems. Furthermore, the findings of this study have broader implications for enhancing societal safety and increasing industrial efficiency.

Being able to effectively read scientific plots, or chart understanding, is a central part toward building effective agents for science. However, existing multimodal large language models (MLLMs), especially open-source ones, are still falling behind with a typical success rate of 30%-50% on challenging benchmarks. Previous studies on fine-tuning MLLMs with synthetic charts are often restricted by their inadequate similarity to the real charts, which could compromise model training and performance on complex real-world charts. In this study, we show that modularizing chart generation and diversifying visual details improves chart understanding capabilities. In particular, we design a five-step data synthesis pipeline, where we separate data and function creation for single plot generation, condition the generation of later subplots on earlier ones for multi-subplot figures, visually diversify the generated figures, filter out low quality data, and finally generate the question-answer (QA) pairs with GPT-4o. This approach allows us to streamline the generation of fine-tuning datasets and introduce the effective chart dataset (ECD), which contains 10k+ chart images and 300k+ QA pairs, covering 25 topics and featuring 250+ chart type combinations with high visual complexity. We show that ECD consistently improves the performance of various MLLMs on a range of real-world and synthetic test sets. Code, data and models are available at: https://github.com/yuweiyang-anu/ECD.

Humans communicate with increasing efficiency in multi-turn interactions, by adapting their language and forming ad-hoc conventions. In contrast, prior work shows that LLMs do not naturally show this behavior. We develop a post-training process to develop this ability through targeted fine-tuning on heuristically identified demonstrations of convention formation. We evaluate with two new benchmarks focused on this capability. First, we design a focused, cognitively-motivated interaction benchmark that consistently elicits strong convention formation trends in humans. Second, we create a new document-grounded reference completion task that reflects in-the-wild convention formation behavior. Our studies show significantly improved convention formation abilities in post-trained LLMs across the two evaluation methods.

We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd overall among all evaluated models and 2nd on agentic benchmarks. We release both GLM-4.5 (355B parameters) and a compact version, GLM-4.5-Air (106B parameters), to advance research in reasoning and agentic AI systems. Code, models, and more information are available at https://github.com/zai-org/GLM-4.5.