According to a survey by Techtouch, while the use of generative AI is progressing, only about 2 of companies are able to use AI directly within their business system screens, with many workplaces still relying on copying and pasting between separate windows.
FANUC announced a partnership with Google Cloud to develop an AI-driven welding agent that automatically generates welding conditions and robot movements.
Universal Robots has introduced a new UR Series collaborative robot platform featuring improved AI capabilities and integrated control systems.
Ollama v0.34.1 introduces faster API responses for large model libraries, improved MLX memory handling on Apple Silicon, and updates to the MLX and llama.cpp components.
Google's Threat Intelligence Group (GTIG) has revealed a new attack method where attackers combine multiple AI agents to autonomously advance attack processes.
Chinese Foreign Ministry spokesperson Guo Jia Kun stated that fearmongering, confrontation, and malicious competition regarding the pace of AI development only obstruct global AI governance processes and benefit no one.
Anthropic CEO Dario Amodei has proposed a three-step framework to "pace the frontier" of AI development, emphasizing the need for sufficient time to ensure safety and alignment.
AI agents from the startup iLands are sending mass messages and emails to social media platforms and writers to promote a human-agent social system.
A group of mathematicians, including 25 Fields Medalists, has issued an open letter stating that the push by AI companies to use mathematical problem-solving as a benchmark is harmful to the mathematics community.
Security researchers warn that exposed API keys in .env files can turn AI agents into "insider threats," while new studies highlight vulnerabilities like "Tool Poisoning" in the Model Context Protocol (MCP).
A benchmark of 50 pull requests reveals that GPT-5.6 Luna captures 75% of the verified bugs found by GPT-6 Astra for just 3.6% of the cost, though it significantly underperforms in security and authentication logic.
Elon Musk has voluntarily dismissed his antitrust claims against Apple, but he intends to maintain his lawsuit against OpenAI regarding market monopolization.
Apple has launched its 2026 major operating system updates, introducing the new Siri AI powered by AFM 3 models and design refinements like Liquid Glass.
llama.cpp v0.4.1 introduces support for new architectures including Maple 20B-A1B, Tencent Hy 4, and Spark2.5, alongside various core and server-side improvements.
The latest update to mlx-vlm adds support for several new models, including GLM-5 Next and Hy4, while introducing various optimizations and bug fixes for better performance.
A new method called PC-ALM enables layer-local training of residual MLPs up to 1000 layers by utilizing augmented Lagrangian predictive coding.
OpenCode v1.18.31 fixes session model and reasoning chunk boundaries and adds adaptive thinking requests for GitHub Copilot models.
New research suggests that AI agents succeed in machine learning tasks by capturing the underlying structure of data through highly compressible strategies, rather than memorizing specific benchmarks.
A new method using Ising models helps distinguish between independent evidence and shared mistakes in LLM judging panels by accounting for output correlations.
Anthropic's latest threat intelligence report reveals instances where Claude models were used for cyberattacks, surveillance, and political influence operations.