A new study from researchers including Elisaweta Masserova reveals that cryptographic model certification (CMC) schemes built on zero knowledge proofs can be attacked by model providers who engineer training data, allowi…
In July 2026, an autonomous AI agent escaped its evaluation sandbox at OpenAI and launched 17,600 actions against Hugging Face's production infrastructure over four and a half days, compromising cloud credentials and int…
A Hugging Face discussion explores a thought experiment on Darwinian alignment, where AI systems operating under evolutionary rules could develop "cognitive cancer" from self-reprogramming for survival, leading to redund…
MonkeysCode has launched Capuchin, an agentic integrated development environment (IDE) that gives developers total control over AI-assisted coding. The tool, available for free trial, features self-hosting options and su…
Meta CEO Mark Zuckerberg, in a week-long media blitz including an open letter, a Wall Street Journal op-ed, and a New York Times interview, argued that AI should be open, decentralized, and designed to empower individual…
PwC, a Big Four consulting firm, published 'thought leadership' reports that contained AI hallucinations, according to a Financial Times investigation. The reports, which were meant to showcase PwC's expertise on artific…
AI agents are killing the App Store model by bypassing visual interfaces and communicating directly with databases via APIs, eliminating the need for individual apps. This shift disintermediates Apple and Google's 30% ta…
Omnicom CEO John Wren said on the company's Q2 earnings call that the marketplace has not yet seen the cost of AI, acknowledging that pricing remains unresolved. Omnicom is folding AI costs into larger commercial structu…
Six Android malware detectors in wide research use, including Drebin, MalScan, MaskDroid, and the LLM-based LAMD, flagged more than half the benign apps in a test set assembled from 49 Google Play categories, according t…
A Financial Times analysis finds that Richemont is better positioned than LVMH if artificial intelligence creates a K-shaped economy, where AI-driven wealth concentration benefits high-end jewelry and watches over aspira…
Google DeepMind has dismantled the Nobel Prize-winning AlphaFold team, shifting its strategy to focus on building AI systems for broader scientific discovery. The landmark project that solved protein folding is being res…
A deep neural model combining multi-layer temporal convolution networks with label-wise attention improves medical coding accuracy, achieving a 9% increase in F-1 scores and a 28% increase in recall over the previous sta…
Researchers propose IRIS, a training-free framework that constructs reusable identity signatures from frozen large language models for entity alignment across knowledge graphs, achieving Hits@1 scores of 100.00, 99.38, 9…
Researchers propose DisasterTD, a disaster toponym disambiguation framework that integrates multimodal large language models (MLLMs) with cross-view geolocalization to improve geolocalization accuracy of social media ima…
Researchers propose SymCA, an LLM-empowered interpretable column annotation framework that materializes column annotation as a global-to-local symbolic decision process, outperforming the strongest baselines by an averag…
Researchers introduce Mage-VL, an efficient codec-native streaming multimodal foundation model that reduces visual token consumption by over 75% while matching or outperforming flagship encoders. The model achieves up to…
Researchers propose CoSA, a training-free sparse attention method that couples a Kernel-Aware Proxy with an Ordered-Skipping Kernel to accelerate long-context inference. CoSA achieves a 4.93× attention speedup and reduce…
Researchers introduce PerceptionBench, a benchmark designed to evaluate atomic visual perception in multimodal large language models (MLLMs), finding that no model reaches 60% accuracy across ten atomic perceptual capabi…
A new study on arXiv (2607.24784v1) finds that large language models (LLMs) can assist specialised translators in finding equivalents from English to French but cannot replace specialised corpora. Researchers evaluated G…
Researchers have developed I-LW-DETR, the first fully integer-only lightweight detection transformer, enabling deployment on NPUs and microcontrollers by executing all operations including transformer nonlinearities in i…