Despite improvements by length extrapolation, efficient attention and memory modules, handling infinitely long documents with linear complexity without performance degradation during extrapolation remains the ultimate challenge in long-text processing. We directly optimize for long-text tasks in an end-to-end fashion and introduce a novel agent workflow, MemAgent, which reads text in segments and updates the memory using an overwrite strategy. We extend the DAPO algorithm to facilitate training via independent-context multi-conversation generation. MemAgent has demonstrated superb long-context capabilities, being able to extrapolate from an 8K context trained on 32K text to a 3.5M QA task with performance loss < 5% and achieves 95%+ in 512K RULER test.

Amid a tidal wave of misinformation flooding social media during elections and crises, extensive research has been conducted on misinformation detection, primarily focusing on text-based or image-based approaches. However, only a few studies have explored multimodal feature combinations, such as integrating text and images for building a classification model to detect misinformation. This study investigates the effectiveness of different multimodal feature combinations, incorporating text, images, and social features using an early fusion approach for the classification model. This study analyzed 1,529 tweets containing both text and images during the COVID-19 pandemic and election periods collected from Twitter (now X). A data enrichment process was applied to extract additional social features, as well as visual features, through techniques such as object detection and optical character recognition (OCR). The results show that combining unsupervised and supervised machine learning models improves classification performance by 15% compared to unimodal models and by 5% compared to bimodal models. Additionally, the study analyzes the propagation patterns of misinformation based on the characteristics of misinformation tweets and the users who disseminate them.

We extend BeamAttack, an adversarial attack algorithm designed to evaluate the robustness of text classification systems through word-level modifications guided by beam search. Our extensions include support for word deletions and the option to skip substitutions, enabling the discovery of minimal modifications that alter model predictions. We also integrate LIME to better prioritize word replacements. Evaluated across multiple datasets and victim models (BiLSTM, BERT, and adversarially trained RoBERTa) within the BODEGA framework, our approach achieves over a 99\% attack success rate while preserving the semantic and lexical similarity of the original texts. Through both quantitative and qualitative analysis, we highlight BeamAttack's effectiveness and its limitations. Our implementation is available at https://github.com/LucK1Y/BeamAttack

Recommendation systems have become essential in modern music streaming platforms, shaping how users discover and engage with songs. One common approach in recommendation systems is collaborative filtering, which suggests content based on the preferences of users with similar listening patterns to the target user. However, this method is less effective on media where interactions are sparse. Music is one such medium, since the average user of a music streaming service will never listen to the vast majority of tracks. Due to this sparsity, there are several challenges that have to be addressed with other methods. This review examines the current state of research in addressing these challenges, with an emphasis on the role of content filtering in mitigating biases inherent in collaborative filtering approaches. We explore various methods of song classification for content filtering, including lyrical analysis using Large Language Models (LLMs) and audio signal processing techniques. Additionally, we discuss the potential conflicts between these different analysis methods and propose avenues for resolving such discrepancies.

Video-to-Audio (V2A) Generation achieves significant progress and plays a crucial role in film and video post-production. However, current methods overlook the cinematic language, a critical component of artistic expression in filmmaking. As a result, their performance deteriorates in scenarios where Foley targets are only partially visible. To address this challenge, we propose a simple self-distillation approach to extend V2A models to cinematic language scenarios. By simulating the cinematic language variations, the student model learns to align the video features of training pairs with the same audio-visual correspondences, enabling it to effectively capture the associations between sounds and partial visual information. Our method not only achieves impressive improvements under partial visibility across all evaluation metrics, but also enhances performance on the large-scale V2A dataset, VGGSound.

Unfairness is a well-known challenge in Recommender Systems (RSs), often resulting in biased outcomes that disadvantage users or items based on attributes such as gender, race, age, or popularity. Although some approaches have started to improve fairness recommendation in offline or static contexts, the issue of unfairness often exacerbates over time, leading to significant problems like the Matthew effect, filter bubbles, and echo chambers. To address these challenges, we proposed a novel framework, Hypergraph Contrastive Multi-Interest Learning for Fair Conversational Recommender System (HyFairCRS), aiming to promote multi-interest diversity fairness in dynamic and interactive Conversational Recommender Systems (CRSs). HyFairCRS first captures a wide range of user interests by establishing diverse hypergraphs through contrastive learning. These interests are then utilized in conversations to generate informative responses and ensure fair item predictions within the dynamic user-system feedback loop. Experiments on two CRS-based datasets show that HyFairCRS achieves a new state-of-the-art performance while effectively alleviating unfairness. Our code is available at https://github.com/zysensmile/HyFairCRS.

The Velvet Sundown's indie ballads, with guitar music and male vocals, is fairly easy, if bland, on the ear. With lyrics such as "eyes like film in faded light, dreams walk barefoot into the night" and "ash and velvet, smoke and flame, calling out in freedom's name", it could all feasibly be either AI-generated or penned by humans. Deezer, a rival music streaming platform, said that its AI detector tool had flagged the music as being "100% AI generated". Spotify did not respond to a request for comment. CEO Daniel Ek has previously told the BBC that he did not intend to ban AI-generated music from the platform but added that he did not agree with using the tech to mimic real artists.

Rep. Marjorie Taylor Greene, R-Ga., joins'Sunday Morning Futures' to discuss whether the government should regulate artificial intelligence, and how AI ties into President Donald Trump's spending bill. As large tech companies continue to take the lead implementing artificial intelligence (AI) into their platforms and workplaces, the latest Fox News national survey finds that while positive reviews of AI have increased, many remain skeptical about its role in society. The survey, released Thursday, finds 43% view AI technology as a good thing for society, up 5 points from April 2023. Still, nearly half of voters, 47%, think AI is bad for society -- about where it was two years ago (46% bad in April 2023). Overall, urban voters (60%), nonwhite voters (56%), voters under age 45 (53%), and men (52%) are those most likely to say AI is a good thing, while rural voters (55%), White voters (51%), voters ages 45 and over (49%), and women (55%) are likely to say it's a bad thing.

Amazon Prime Day 2025 is fast approaching, and the sale is already underway on some items. To help you find the best early Prime Day deals, we've scoured Amazon for deals on the tech we love. As always, every deal we recommend here is on a product our reviewers have personally tested and approved--you won't find any shoddy dupes or mystery brands here. This year Prime Day runs for four days, July 8-11, rather than the usual two. That means there's twice as long to suffer save. Be sure to read our explainer on all the Amazon Prime perks you should be taking advantage of. Updated Thursday, July 3, 2025: We've add deals on Amazon's Kindle Essentials Bundle, Echo Spot, an Arlo security cam, two Tapo cams, the Jackery Explorer 300 power station, the Glimpse Sleep Mask, Brooklinen's organic sheets, and more. If you're looking to get a new Kindle and want a case, then snag this handy essentials kit while it's on sale for Prime Day. It includes the latest basic Kindle, a fabric cover, and a power adapter (which is also handy since Kindles only come with a charging cord, no adapter). The bundle only comes with a black Kindle, but you can choose from a couple of cover colors.

AI slop is flooding every single digital platform, and music streaming services are no exception--so much so, even someone who generally avoids AI might find themselves unknowingly listening to a robot hornily singing about butts. Take the sordid saga of "Make Love to My Shitter," an AI-generated track from an artist called BannedVinylCollection. Brace Belden, a host of the popular politics podcast TrueAnon, says that Spotify recently queued up the bawdy song after he'd finished listening to alt-country legend Lucinda Williams' 1992 album Sweet Old World. "I didn't realize the song was AI at first," he says. "I thought it might've been some obscene joke record from the 80s or 90s."