Membership in ACM includes a subscription to Communications of the ACM (CACM), the computing industry's most trusted source for staying connected to the world of advanced computing. In an era defined by rapid technological advancement, particularly in fields such as artificial intelligence (AI), there is a growing discourse surrounding the pivotal role of academia and the impact of federal funding on innovation. The following conversation sheds light on an often-underdiscussed facet of this relationship: the profound influence of academic research on the formation and continued success of large technology companies such as Google. The participants include Magda Balazińska (MB) and three senior Google engineers--Urs Hölzle (UH), Jeff Dean (JD), and Parthasarathy Ranganathan (PR)--who collectively have more than a century of experience spanning both academia and industry, and between them represent different disciplines across the computing stack (distributed systems, AI, hardware). The discussion delves into the foundational role of academia in Google's inception, the long-term impact of federally funded research, the stories behind key innovations, and the grand challenges that lie ahead for academic research.

Talks in Geneva between the US and Ukraine aimed at ending the war with Russia have concluded, with officials from both sides reporting progress and an intention to continue working. However, no details have emerged on how to bridge the considerable divide between Moscow and Kyiv over territorial issues and security guarantees for Ukraine. Ukraine's president Volodymyr Zelensky welcomed the important steps that had been made but warned that the main problem facing the peace talks was Vladimir Putin's demand for legal recognition of Russian-occupied territories in eastern Ukraine. This would break the principle of territorial integrity and sovereignty, he said, highlighting concerns that Moscow could be rewarded for its aggression with land it seized by force. Meanwhile, President Donald Trump suggested on social media that something good just may be happening, but with the caveat: Don't believe it until you see it.

For months Rachel Reeves has been schmoozing the biggest players in the UK government debt market to ensure the smooth passage of her plans. For months Rachel Reeves has been schmoozing the biggest players in the UK government debt market to ensure the smooth passage of her plans. Bond market power: why Rachel Reeves is keen to keep the £2.7tn'beast' onside Hugely influential traders will be hanging on the chancellor's every word when she announces her budget At just after 12.30pm on Wednesday, the machine will be listening, the trading algorithms ready, and billions of pounds of buy-and-sell orders stacked up awaiting Rachel Reeves's budget. For the first time on the London trading floor of Deutsche Bank, a custom-built artificial intelligence tool will tune in to the chancellor's speech. It will transcribe her words, spot shifts in tone and spit out alerts when the numbers deviate from expectations.

You live in a house you designed and built yourself. You rely on the sun for power, heat your home with a woodstove, and farm your own fish and vegetables. This is the life of Marcin Jakubowski, the 53-year-old founder of Open Source Ecology, an open collaborative of engineers, producers, and builders developing what they call the Global Village Construction Set (GVCS). It's a set of 50 machines--everything from a tractor to an oven to a circuit maker--that are capable of building civilization from scratch and can be reconfigured however you see fit. It's all part of his ethos that life-changing technology should be available to all, not controlled by a select few. What it's like to find yourself in the middle of a conspiracy theory Last week, we held a subscribers-only Roundtables discussion exploring how to cope in this new age of conspiracy theories.

Poisoning is a term most often associated with the human body and natural environments . But it is also a growing problem in the world of artificial intelligence (AI) - in particular, for large language models such as ChatGPT and Claude. In fact, a joint study by the UK AI Security Institute, Alan Turing Institute and Anthropic, published earlier this month, found that inserting as few as 250 malicious files into the millions in a model's training data can secretly "poison" it. So what exactly is AI poisoning? And what risks does it pose?

'No one will support it': Ukraine's soldiers react to US peace plan Ukraine's frontline soldiers have reacted to draft US peace proposals with a mixture of defiance, anger and resignation. The BBC spoke to half a dozen who sent us their views via social media and email in response to the original US plan - details of which were leaked last week. Since then, American and Ukrainian negotiators have been working on changes to the proposals - and are set to continue talks about the peace framework. Of the original US plan, Yaroslav, in eastern Ukraine, says it sucks no one will support it while an army medic with the call sign Shtutser dismissed it as an absolutely disgraceful draft of a peace plan, unworthy of our attention. But one soldier with the call sign Snake told us it's time to agree at least on something.

The report found 7% of respondents had been depicted in a sexual or intimate deepfake. The report found 7% of respondents had been depicted in a sexual or intimate deepfake. One in four people think there is nothing wrong with creating and sharing sexual deepfakes, or they feel neutral about it, even when the person depicted has not consented, according to a police-commissioned survey. The findings prompted a senior police officer to warn that the use of AI is accelerating an epidemic in violence against women and girls (VAWG), and that technology companies are complicit in this abuse. The survey of 1,700 people commissioned by the office of the police chief scientific adviser found 13% felt there was nothing wrong with creating and sharing sexual or intimate deepfakes - digitally altered content made using AI without consent.

Single-cell RNA sequencing (scRNA-seq) enables transcrip-tomic profiling at cellular resolution but suffers from pervasive dropout events that obscure biological signals. We present SCR-MF, a modular two-stage workflow that combines principled dropout detection using scRecover with robust non-parametric imputation via missForest. Across public and simulated datasets, SCR-MF achieves robust and interpretable performance comparable to or exceeding existing imputation methods in most cases, while preserving biological fidelity and transparency. Runtime analysis demonstrates that SCR-MF provides a competitive balance between accuracy and computational efficiency, making it suitable for mid-scale single-cell datasets.

Ensuring character identity consistency across varying prompts remains a fundamental limitation in diffusion-based text-to-image generation. We propose CharCom, a modular and parameter-efficient framework that achieves character-consistent story illustration through composable LoRA adapters, enabling efficient per-character customization without retraining the base model. Built on a frozen diffusion backbone, CharCom dynamically composes adapters at inference using prompt-aware control. Experiments on multi-scene narratives demonstrate that CharCom significantly enhances character fidelity, semantic alignment, and temporal coherence. It remains robust in crowded scenes and enables scalable multi-character generation with minimal overhead, making it well-suited for real-world applications such as story illustration and animation.

Adversarial attacks on tabular data present unique challenges due to the heterogeneous nature of mixed categorical and numerical features. Unlike images where pixel perturbations maintain visual similarity, tabular data lacks intuitive similarity metrics, making it difficult to define imperceptible modifications. Additionally, traditional gradient-based methods prioritise $\ell_p$-norm constraints, often producing adversarial examples that deviate from the original data distributions. To address this, we propose a latent-space perturbation framework using a mixed-input Variational Autoencoder (VAE) to generate statistically consistent adversarial examples. The proposed VAE integrates categorical embeddings and numerical features into a unified latent manifold, enabling perturbations that preserve statistical consistency. We introduce In-Distribution Success Rate (IDSR) to jointly evaluate attack effectiveness and distributional alignment. Evaluation across six publicly available datasets and three model architectures demonstrates that our method achieves substantially lower outlier rates and more consistent performance compared to traditional input-space attacks and other VAE-based methods adapted from image domain approaches, achieving substantially lower outlier rates and higher IDSR across six datasets and three model architectures. Our comprehensive analyses of hyperparameter sensitivity, sparsity control, and generative architecture demonstrate that the effectiveness of VAE-based attacks depends strongly on reconstruction quality and the availability of sufficient training data. When these conditions are met, the proposed framework achieves superior practical utility and stability compared with input-space methods. This work underscores the importance of maintaining on-manifold perturbations for generating realistic and robust adversarial examples in tabular domains.