Academics discover black people'significantly more likely' to be identified when compared with other ethnic groups Essex police have paused the use of live facial recognition (LFR) technology after a study found cameras were significantly more likely to target black people than people of other ethnicities. The move to suspend use of the AI-enabled systems was revealed by the Information Commissioner's Office (ICO), which regulates the use of the technology deployed so far by at least 13 police forces in London, south and north Wales, Leicestershire, Northamptonshire, Hampshire, Bedfordshire, Suffolk, Greater Manchester, West Yorkshire, Surrey and Sussex. The ICO said Essex police had paused LFR deployments "after identifying potential accuracy and bias risks" and warned other forces to have mitigations in place. LFR systems are either mounted to fixed locations or deployed in vans. In January, the home secretary, Shabana Mahmood, announced the number of LFR vans would increase five-fold, with 50 available to every police force in England and Wales. Essex commissioned University of Cambridge academics to conduct a study, which involved 188 actors walking past cameras being actively deployed from marked police vans in Chelmsford.

PCWorld reports that Microsoft's Clipchamp video editor in Windows 11 now mandates OneDrive for saving and editing video projects. This change significantly impacts users who prefer local storage, as locally saved projects become uneditable archives that cannot be modified. New Clipchamp projects automatically sync to OneDrive accounts, though media files within projects may not always require cloud synchronization. Microsoft is changing how Clipchamp--the built-in free video editor for Windows 11--works. The program now requires video projects to be saved to Microsoft's OneDrive cloud storage service in order to continue editing them, reports Windows Latest .

Video game fans and big, blockbuster releases have had an uneasy relationship in recent years. As so-called triple-A games get more expensive to make, the publishers behind them are accused of taking fewer risks and failing to try new things. But highly anticipated new release Crimson Desert asks a different question - what if a big-budget, graphically advanced game tried to do absolutely everything? The ambitious action-adventure's been compared to a buffet, presenting players with a smorgasbord of ideas, gameplay styles and quests to gorge on. While some have praised it as a feast, others have found it overstuffed, with some undercooked morsels behind the impressive presentation.

Instant to reduce annoying "if you want" and teaser-style phrasing that users found intrusive. This change addresses widespread user complaints about persistent, clickbait-like follow-up prompts that negatively impacted the AI interaction experience. The update aims to create more natural, direct conversations by making ChatGPT less chatty and eliminating the bothersome response teasers. It wasn't all that long ago that ChatGPT was a constant nag, persistently dropping "Would you like me to?"-style questions at the end of its responses. OpenAI eventually tweaked the phrasing, dropping the question marks and going for "if you want"-style teasers that invited users to extend their chat sessions. Now, OpenAI has acknowledged that it went too far with the clickbaity follow-ups, noting in a recent update for one of its newest models that it's now cutting back on the teasers. "We're rolling out an update to GPT-5.3 Instant that improves follow-up tone and reduces teaser-style phrasing," reads a recent ChatGPT release note, which adds that users should soon see fewer follow-ups like "if you want," "you'll never believe," and "I can tell you three things that " Those teasers are, of course, a way for ChatGPT to keep subscribers chatting, but users have been complaining that the persistent follow-ups are more annoying than they are intriguing. "I hated it with a passion and hope it's completely gone," wrote one user on Reddit .

'Retro Rewind' can make it a Blockbuster night. The two-person development team says "nostalgia is a central element" of their 90s themed simulator. Breakthroughs, discoveries, and DIY tips sent six days a week. Growing up in the early 2000s, few weekly rituals stuck with me quite like New Release Tuesday. Every week, without fail, I remember wandering the slightly-moldy-smelling, blue-carpeted aisles of our local Blockbuster while my mom scrutinized the newest covers.

Amazon's next-generation voice assistant launches in early access in Europe for the first time. Amazon's next-generation smart assistant has entered its Early Access program in the UK, marking Alexa+'s European debut following rollouts in the US, Canada and Mexico. Starting March 19, invitations to start using the smarter, more conversational will be sent out to hundreds of thousands of willing participants, Amazon said in a, adding that Alexa is the most popular voice assistant in the UK. As well as its more natural communication, agentic capabilities, contextual awareness and ability to remember previous conversations across devices, Amazon that users across the pond are getting an authentically British AI-powered assistant. It understands slang terms like cuppa and might even accuse you of taking the mick in the middle of a conversation.

Signal's Creator Is Helping Encrypt Meta AI Moxie Marlinspike says the technology powering his encrypted AI chatbot, Confer, will be integrated into Meta AI. The move could help protect the AI conversations of millions of people. Moxie Marlinspike, cofounder of the Signal Foundation, says his new privacy-focused AI platform, Confer, will be integrated into Meta AI. Moxie Marlinspike, the privacy advocate who created the secure communication app Signal and its widely used open source encryption protocol, said this week that his privacy-focused AI platform, Confer, will start incorporating its technology into Meta's AI systems. Every day, billions of chat messages sent through Signal, Meta's WhatsApp, and Apple's Messages are protected by end-to-end encryption .

Fine-tuning is a crucial process for adapting large language models (LLMs) to diverse applications. In certain scenarios, such as multi-tenant serving, deploying multiple LLMs becomes necessary to meet complex demands. Recent studies suggest decomposing a fine-tuned LLM into a base model and corresponding delta weights, which are then compressed using low-rank or low-bit approaches to reduce costs. In this work, we observe that existing low-rank and low-bit compression methods can significantly harm the model performance for task-specific fine-tuned LLMs (e.g., WizardMath for math problems). Motivated by the long-tail distribution of singular values in the delta weights, we propose a delta quantization approach using mixed-precision. This method employs higher-bit representation for singular vectors corresponding to larger singular values. We evaluate our approach on various fine-tuned LLMs, including math LLMs, code LLMs, chat LLMs, and even VLMs. Experimental results demonstrate that our approach performs comparably to full fine-tuned LLMs, surpassing both low-rank and low-bit baselines by a considerable margin. Additionally, we show that our method is compatible with various backbone LLMs, such as Llama-2, Llama-3, and Mistral, highlighting its generalizability.

Kernel methods are widely utilized in machine learning field to learn, from training data, a latent function in a reproducing kernel Hilbert space. It is well known that the approximator thus obtained usually achieves a linear representation, which brings various computational benefits, while maintaining great representation power (i.e., universal approximation). However, when non-negativity constraints are imposed on the function's outputs, the literature usually takes the kernel method-based approximators as offering linear representations at the expense of limited model flexibility or good representation power by allowing for their nonlinear forms. The main contribution of this paper is to derive a sufficient condition for a positive definite kernel so that it may construct flexible and linear approximators of non-negative functions. We call a kernel function that offers these attributes an; it is reminiscent of the inverse M-matrix. Furthermore, we show that for a one-dimensional input space, universal exponential/Abel kernels are inverse M-kernels and construct linear universal approximators of non-negative functions. To the best of our knowledge, it is the first time that the existence of linear universal approximators of non-negative functions has been elucidated. We confirm the effectiveness of our results by experiments on the problems of non-negativity-constrained regression, density estimation, and intensity estimation. Finally, we discuss issues and perspectives on multi-dimensional input settings.

Modern image generation (IG) models have been shown to capture rich semantics valuable for image understanding (IU) tasks. However, the potential of IU models to improve IG performance remains uncharted. We address this issue using a token-based IG framework, which relies on effective tokenizers to project images into token sequences. Currently, **pixel reconstruction** (e.g., VQGAN) dominates the training objective for image tokenizers. In contrast, our approach adopts the **feature reconstruction** objective, where tokenizers are trained by distilling knowledge from pretrained IU encoders.