6
“新奇机器人”完成A+轮融资
Robotics startup Xinqi Robot has completed its third funding round in six months, attracting investment from both venture capital firms and industrial partners.
6
全网骂Claude变笨，Anthropic下场揭秘：坑你的不是模型
Anthropic clarified that recent developer complaints about Claude Code underperforming stem from user confusion between model selection and the effort setting rather than actual model degradation.
6
US Air Force Research Lab to relocate TI-19 supercomputer from Ohio to Mississippi
The US Air Force Research Laboratory is relocating its 47-petaflops TI-19 supercomputer from Ohio to Mississippi to extend its operational lifespan.
6
CleanSpark secures tenant with 20-year lease for data center in Sandersville, Georgia
CleanSpark has signed a twenty-year lease agreement for a data center facility in Georgia valued at 6.6 billion dollars in total revenue.
6
PHANES AI Writes Touch Into Robot Foundation Models: TouchWorld Tactile Model Enables Dexterous Manipulation
PHANES AI has released TouchWorld, a predictive and reactive tactile foundation model designed to give robots a sense of touch for precise physical tasks.
6
ICML 2026世界模型研究盘点：LAWM VS WAM，谁主沉浮？
An analysis of ICML 2026 papers reveals a shifting paradigm in world model research, with a strong resurgence in World Action Models compared to Latent Action World Models.
6
Stripe与Advent据悉联合提出530亿美元收购PayPal
Stripe and private equity firm Advent International have reportedly made a joint 53 billion dollar bid to acquire PayPal.
6
The AI Arms Race in Technical Interviews Is Escalating
The technical hiring process is experiencing an escalation as job applicants use real-time AI assistants and employers counter with AI-powered detection tools.
6
Security complexity in data centers: when more controls do not necessarily mean more control
This article discusses the challenges of managing security complexity within data center environments.
6
The real AI race may no longer be at the frontier
Hugging Face CEO Clem Delangue argues that enterprise demand is shifting toward open-source models due to advantages in cost and data ownership.
6
Unitree Robotics Partners With Hunan Steel: A Physical AI Infrastructure Company Emerges
Unitree Robotics has partnered with Hunan Steel and Looper Robotics to deploy legged robots for industrial inspection, signaling a move toward physical AI in heavy manufacturing.
6
Chinese AI labs to challenge Thinking Machines Lab with new industry focus
Former Chinese AI lab leaders are launching a new venture focused on industry-specific AI solutions to compete with Thinking Machines Lab by prioritizing practical applications over general-purpose models.
6
逐际动力再融2亿美元提速港股IPO，张巍：营收对赌不符合具身商业逻辑
LimX Dynamics has raised $200 million in new funding as it prepares for a potential IPO while focusing on the commercial viability of embodied AI robotics.
6
Silent Alarm: A J-Space Protocol for Comparing Danger Recognition Across Models and Quantization Levels
A new protocol called JADR has been proposed to measure the underlying safety fragility of language models beyond standard jailbreak testing.
6
Mistake gating leads to energy and memory efficient continual learning
A biologically inspired learning method called mistake-gated learning is introduced to reduce the energy and memory requirements of training neural networks.
6
Attractor Geometry of Transformer Memory: From Conflict Arbitration to Confident Hallucination
Researchers identified a unified geometric mechanism in transformer models that explains how conflicting memory sources lead to confident hallucinations.
6
DRIFTLENS: Measuring Memory-Induced Reasoning Drift in Personalized Language Models
The study introduces DRIFTLENS to measure how personalized memory injection in LLMs can alter the reasoning trajectories used to generate responses.
6
dMX: Differentiable Mixed-Precision Assignment for Low-Precision Floating-Point Formats
The dMX framework introduces a differentiable approach to mixed-precision quantization, optimizing LLM performance and efficiency for low-precision hardware.
6
Inclusive Federated Learning Through Compliance-Weighted Noise Allocation in Healthcare AI
Researchers proposed a compliance-aware federated learning framework that adjusts differential privacy noise to accommodate varying institutional data standards and resource levels.
6
Tracking Drift: Variation-Aware Entropy Scheduling for Non-Stationary Reinforcement Learning
Researchers have developed a variation-aware entropy scheduling method to improve reinforcement learning performance in environments subject to drift.
6
TADPO: Reinforcement Learning Goes Off-road
Researchers have introduced TADPO, a reinforcement learning approach designed to improve autonomous vehicle navigation in complex, unmapped off-road environments.
6
Comparing Semantic Navigation in Humans and Large Language Models using Natural Language Processing
A comparative study using NLP metrics reveals similarities and differences in how humans and leading large language models navigate conceptual spaces during semantic memory retrieval.
6
Are we Merging the Right Models? Impact of Expert Training Duration on Model Merging for LLMs
The authors investigate how the training duration of individual domain experts influences the performance of merged large language models.
6
Plug Power sells Texas site to Stream Data Centers
Plug Power has sold a Texas industrial site to Stream Data Centers, which intends to repurpose the land for data center infrastructure.