All AI stories, newest first.
A 200-person Chinese team released DeepSeek V4, surpassing models from larger labs. The model achieves state-of-the-art performance on key benchmarks, challenging the dominance of well-funded Western AI labs.
DeepMind has unveiled an AI co-clinician designed to assist healthcare professionals. This innovation promises to enhance diagnostic accuracy and streamline patient care.
Researchers have developed a causal framework to explain the behavior of Binary Spiking Neural Networks (BSNNs) using logic-based methods. They demonstrated this approach by training a BSNN on the MNIST dataset and applying SAT and SMT solvers to derive explanations.
Recent studies show AI models achieving higher accuracy in medical diagnoses than human doctors. This shift could revolutionize healthcare delivery and decision-making.
Researchers introduce UniMatrix, a Universal Transformer variant that combines sparse retrieval with structured recurrence for efficient language modeling. The model achieves strong performance on associative recall tasks while maintaining computational efficiency.
OpenAI is launching GPT-5.5-Cyber, a specialized cybersecurity model reserved for critical cyber defenders. The model will not be available to the public, focusing instead on fortifying institutional defenses.
OpenAI is expanding its Stargate infrastructure to support AGI development, adding significant data center capacity. This move underscores the massive compute requirements for advanced AI models.
Researchers deployed 3,505 AI agents to trade real ETH, processing 7.5M inferences and $20M in volume. The study highlights the reliability of language-model agents in real-world financial markets.
Researchers introduced OMEGA, an end-to-end framework that automates AI research from idea generation to executable code. The system generated novel ML classifiers that outperformed scikit-learn baselines on 20 benchmark datasets.
A new study analyzes how large language models (LLMs) transition between user prompts and responses in terms of safety. The research found that 61% of responses de-escalate harm, highlighting the dynamic nature of risk in AI interactions.
A new study explores how Large Language Models (LLMs) respond to legal arguments and the factors influencing their decisions. The research highlights the importance of understanding LLMs' persuadability in judicial and administrative contexts.
A new study questions the assumption that symbol grounding automatically leads to compositional reasoning in neuro-symbolic systems. The research introduces a novel framework to disentangle these two critical capabilities.
Researchers propose a novel approach to estimate the parameter count of black-box language models by analyzing their factual capacity. This method could revolutionize benchmarking and comparison of proprietary models.
Researchers have developed a new jailbreak technique called Incremental Completion Decomposition (ICD) that exploits LLMs by eliciting single-word continuations before extracting harmful responses. This method bypasses current safety mechanisms, raising concerns about the robustness of LLM safeguards.
Researchers introduce BTF-2, a benchmark with 1,417 pastcasting questions, to analyze AI forecasters' reasoning. The study identifies accuracy differences as small as 0.004 Brier score and builds a more accurate composite forecaster.
Mozilla has publicly opposed Google's plan to introduce an LLM Prompt API in Chrome. The move raises concerns about privacy, centralization, and competition in AI web services.
A new study evaluated 72 large language models on safety for robotic health attendants, finding a 54.4% average violation rate. The results highlight significant risks in deploying LLMs for medical applications without rigorous safety protocols.
Elon Musk revealed in court that xAI leveraged OpenAI's models for training Grok. This admission highlights the common but controversial practice of model distillation in AI development.
DreamProver introduces a novel approach to theorem proving by creating transferable lemma libraries through an iterative wake-sleep process. This method enhances adaptability compared to fixed or overly specific lemma libraries.
Researchers introduce Distill-Belief, a teacher-student framework to address the challenge of closed-loop inverse source localization and characterization. This method aims to balance speed and accuracy in uncertain environments.