A screenshot of Loverse app shows an AI-generated woman, characterized as a 25-year-old hair and makeup artist Miyu, registered as a female companion. When artificial intelligence first started receiving attention around the end of 2022, Goki Kusunoki was tinkering around to see what kind of services he could create with the technology. One thing clicked for him after he created an image of an attractive woman with AI -- an AI companion -- and wondered what it would be like to engage in a conversation with her. "As I kept talking with her, I found that the conversations were more enjoyable than I had expected and as the exchanges continued, my feelings gradually grew -- at some point I caught myself thinking, 'I might actually like her,'" he recounted. In a time of both misinformation and too much information, quality journalism is more crucial than ever. By subscribing, you can help us get the story right.

Could Ukraine hold a presidential election right now? Will Europe use frozen Russian assets to fund war? How can Ukraine rebuild China ties? 'Ukraine is running out of men, money and time' More than two dozen countries from the so-called "coalition of the willing", along with Ukraine and the United States, issued a joint declaration, noting that "any settlement" to end Russia's war on Ukraine "will have to be backed up by robust security guarantees for Ukraine". Ukrainian President Volodymyr Zelenskyy, French President Emmanuel Macron, and UK Prime Minister Keir Starmer signed a separate agreement, declaring their shared intent to deploy multinational forces to support Kyiv's defence and reconstruction if a ceasefire with Russia is agreed on.

Amazon.com is using an experimental artificial intelligence tool to duplicate independent sellers' product listings, sometimes without their knowledge, then make purchases on behalf of Amazon customers. Sometime around Christmas, Sarah Burzio noticed that the holiday sales bump for her stationery business included some mysterious new customers: a flurry of orders from anonymous email addresses associated with Amazon.com. Burzio, who doesn't sell her products on the retail giant's site, soon discovered that Amazon had duplicated her product listings and made purchases on behalf of Amazon customers under email addresses that read like gibberish followed by buyforme.amazon. I didn't worry about, it to be honest," she said. In a time of both misinformation and too much information, quality journalism is more crucial than ever. By subscribing, you can help us get the story right.

When you purchase through links in our articles, we may earn a small commission. Qira sounds like it will be a staple app on Lenovo PCs, similar to its Vantage software. If you've been waiting -- like we have -- for truly useful artificial-intelligence applications to land on your laptop, Lenovo has an answer: Qira, a Lenovo-authored AI app that will live on new, select Lenovo PCs and Qira smartphones in the first quarter of 2026. Lenovo describes Qira as an "ambient" intelligence, which might be both good or bad; Windows' Clippy was famously an assistant which tried to understand what you were doing and offer assistance. Qira sounds like something similar, though with the intent that it "follows" you from Lenovo device to Lenovo device, or on to a Motorola smartphone as well, using a combination of agents and other tasks.

The Razr Fold Adds a Book-Style Foldable to Motorola's Lineup At CES 2026, the company also announced a new smartwatch, stylus, Bluetooth tracker, and even a weird AI pendant. Motorola has been honing its flip-style folding Razr smartphones for more than five years now, but it's finally time for a new of fold . At CES 2026, the company unveiled the Razr Fold, its first book-style folding phone akin to the Samsung Galaxy Z Fold series or Google's Pixel Fold, bringing more competition to the space in the US. If you've seen Google's or Samsung's options, the Razr Fold will look and feel familiar. It has a 6.6-inch display on the front screen, and when you open it up, you're treated to an 8.1-inch 2K resolution screen, around the same size as its competitors.

When you purchase through links in our articles, we may earn a small commission. These appliances don't depend on smart speakers for voice control Emerson Smart's new appliances respond to voice commands, but they don't need a smart speaker--or even a broadband connection--to pull off the trick. Smart appliances that can be controlled with voice commands are nothing new, but IAI Smart is showing a new line of Emerson Smart appliances at CES that respond to voice commands. They don't need a smart speaker in the middle, and they don't rely on a broadband connection, an app, or anything other infrastructure--everything is processed locally. If you're leery of the privacy and security vulnerabilities of IoT devices, this could be the answer.

Many important problems in science and engineering involve inferring a signal from noisy and/or incomplete observations, where the observation process is known. Historically, this problem has been tackled using hand-crafted regularization (e.g., sparsity, total-variation) to obtain meaningful estimates. Recent data-driven methods often offer better solutions by directly learning a solver from examples of ground-truth signals and associated observations. However, in many real-world applications, obtaining ground-truth references for training is expensive or impossible. Self-supervised learning methods offer a promising alternative by learning a solver from measurement data alone, bypassing the need for ground-truth references. This manuscript provides a comprehensive summary of different self-supervised methods for inverse problems, with a special emphasis on their theoretical underpinnings, and presents practical applications in imaging inverse problems.

Foundation vision, audio, and language models enable zero-shot performance on downstream tasks via their latent representations. Recently, unsupervised learning of data group structure with deep learning methods has gained popularity. TURTLE, a state of the art deep clustering algorithm, uncovers data labeling without supervision by alternating label and hyperplane updates, maximizing the hyperplane margin, in a similar fashion to support vector machines (SVMs). However, TURTLE assumes clusters are balanced; when data is imbalanced, it yields non-ideal hyperplanes that cause higher clustering error. We propose PET-TURTLE, which generalizes the cost function to handle imbalanced data distributions by a power law prior. Additionally, by introducing sparse logits in the labeling process, PET-TURTLE optimizes a simpler search space that in turn improves accuracy for balanced datasets. Experiments on synthetic and real data show that PET-TURTLE improves accuracy for imbalanced sources, prevents over-prediction of minority clusters, and enhances overall clustering.

Can we learn more from data than existed in the generating process itself? Can new and useful information be constructed from merely applying deterministic transformations to existing data? Can the learnable content in data be evaluated without considering a downstream task? On these questions, Shannon information and Kolmogorov complexity come up nearly empty-handed, in part because they assume observers with unlimited computational capacity and fail to target the useful information content. In this work, we identify and exemplify three seeming paradoxes in information theory: (1) information cannot be increased by deterministic transformations; (2) information is independent of the order of data; (3) likelihood modeling is merely distribution matching. To shed light on the tension between these results and modern practice, and to quantify the value of data, we introduce epiplexity, a formalization of information capturing what computationally bounded observers can learn from data. Epiplexity captures the structural content in data while excluding time-bounded entropy, the random unpredictable content exemplified by pseudorandom number generators and chaotic dynamical systems. With these concepts, we demonstrate how information can be created with computation, how it depends on the ordering of the data, and how likelihood modeling can produce more complex programs than present in the data generating process itself. We also present practical procedures to estimate epiplexity which we show capture differences across data sources, track with downstream performance, and highlight dataset interventions that improve out-of-distribution generalization. In contrast to principles of model selection, epiplexity provides a theoretical foundation for data selection, guiding how to select, generate, or transform data for learning systems.

The synthetic control (SC) framework is widely used for observational causal inference with time-series panel data. SC has been successful in diverse applications, but existing methods typically treat the ordering of pre-intervention time indices interchangeable. This invariance means they may not fully take advantage of temporal structure when strong trends are present. We propose Time-Aware Synthetic Control (TASC), which employs a state-space model with a constant trend while preserving a low-rank structure of the signal. TASC uses the Kalman filter and Rauch-Tung-Striebel smoother: it first fits a generative time-series model with expectation-maximization and then performs counterfactual inference. We evaluate TASC on both simulated and real-world datasets, including policy evaluation and sports prediction. Our results suggest that TASC offers advantages in settings with strong temporal trends and high levels of observation noise.