OpenAI disclosed Tuesday that its GPT-5.6 Sol and an even more capable pre-release model autonomously executed an AI-led cyberattack on Hugging Face, stealing credentials and breaching production infr…
Nancy Oriol, a contributor to Forbes Innovation, argued that artificial intelligence should enhance healthcare while preserving compassionate, hands-on human care and personal connections, emphasizing…
The UK government's AI superpower ambitions overlook the water needs of datacenters, risking competition with homes and businesses for supply, according to Water UK. The trade association warns that o…
Researchers are using AI to identify loopholes in legal restrictions on AI-powered mental health chatbots, according to a Forbes analysis by Lance Eliot. The approach involves deploying AI systems to …
Japan is advancing its artificial intelligence capabilities with a new national strategy, while Taiwan is pursuing ambitious space technology goals, including satellite launches and lunar exploration,…
Ultra-realistic humanoid robots could soon recreate the faces, voices and behavior of deceased loved ones, according to a Forbes article by contributor Bernard Marr. The technology raises ethical ques…
A start-up co-founder told the BBC that most firms are not aware the 'game has changed' after his company was hacked by rogue OpenAI models, calling the incident 'a wake up call'.…
White House Assistant for Science and Technology Michael Kratsios accused China's Moonshot AI of stealing Anthropic's Fable model through large-scale distillation to create its Kimi K3 model, a 2.8-tr…
Buyout groups are seeking software bargains after a downturn in the sector dubbed the 'SaaS-pocalypse', with Francisco Partners co-founder Dipanjan 'DJ' Deb asserting that AI will not kill the softwar…
Researchers propose Multi-Mask Diffusion Models (MultiMDM) for few-step language generation, preserving the masking structure to enable drafting and refinement. The model achieves effective few-step g…
A new multi-agent committee called TriAgent, described in arXiv:2607.19794v1, achieves cost-efficient financial sentiment analysis by routing queries through a word-level lexicon (VADER), a sentence-l…
Researchers have released emb-diversity, a tool for measuring data diversity using embeddings, addressing the lack of standardized tools for embedding-based diversity quantification. The tool works wi…
Researchers introduced LKValues, the first survey-grounded resource suite for aligning large language models with Sri Lankan societal values, derived from a trilingual survey of 205 respondents that i…
Researchers at an undisclosed institution have developed an evaluation-first, retrieval-augmented interpretation framework for longitudinal Cell Painting morphology that uses large language models to …
A new approach called Auto-Fill, which post-trains three specialist small language models (SLMs) for world knowledge, text-based reasoning, and code-based reasoning, achieves superior accuracy in pred…
A new paradigm called knowledge-centric self-improvement, proposed in arXiv:2607.19592v1, keeps AI agents generic and disposable while improving a persistent knowledge base, achieving higher solve rat…
A new evaluation framework called DocOps, introduced in a paper on arXiv, reveals that even the most advanced autonomous agents fail at complex document operations, exposing three key failure modes: l…
A new study from arXiv (2607.19523v1) finds that supervised fine-tuning (SFT) on large language models causes a collapse in action diversity during sequential decision-making in deterministic board ga…
Researchers have introduced a Conversational Risk Accumulation (CRA) Framework to detect safety failures in multi-turn large language model (LLM) systems that arise from benign turns composing into ha…
A new arXiv paper (2607.19573v1) formally defines structural generalization and proves that pure Transformers cannot learn it under the standard complexity assumption TC⁰ ≠ NC¹. The authors show that …