Shepherd launches an alpha runtime that makes AI agent executions reversible using Git-like forking, delivering 5x faster copy-on-write performance than Docker commit, 95% KV-cache reuse on replay, and OS-level sandboxed permissions that block unauthorized writes before they happen.
European AI startup UMA unveils a humanoid robot powered by Real-Time Learning architecture that watches and learns from human demonstrations, positioning Europe as a serious contender in the global robotics race against the US and China.
A new multi-agent framework called T3MP3ST is transforming AI coding tools like Claude Code and Codex into autonomous hacking systems capable of running full cyberattack chains, achieving a record-breaking 90.1% score on a leading industry benchmark with fully reproducible results.
AI agents are rapidly hitting production environments, with over half of developers reporting adoption, as Slack emerges as the central coordination hub for multi-agent systems by unifying conversational history, workflow activity, and permissions into a single architecture that reduces risk and overhead for both workers and developers.
Anthropic's Agent Skills standard is taking the AI world by storm, amassing 22,500 GitHub stars as its open, portable format for packaging specialized AI capabilities gains rapid adoption across a growing ecosystem of tools and agentic clients.
Woodside Energy is deploying 50 AI agents across its operations, scaling toward a fully autonomous enterprise by combining decades of industrial data with agentic AI tools, strict governance, and a 'think big, prototype small, scale fast' strategy to optimize LNG operations and reduce energy costs.
Google rolls out Gemini Spark to macOS, enabling AI-powered local file automation, folder organization, and document building on desktop — with remote triggering from a phone coming soon — but access requires a Google AI Ultra subscription starting at $99 per month.
Apple Silicon is emerging as the dominant force in on-device AI, with frontier labs ditching the cloud in favor of Mac minis and Mac Studios to run powerful AI agents 24/7, thanks to Apple's Neural Engine, unified memory, and unmatched power efficiency driving a major shift toward local AI inference.
VulnClaw launches as a powerful open-source AI penetration testing tool that automates the full attack pipeline from reconnaissance to report generation, featuring anti-hallucination gates that reject unverified results, support for 13 LLM providers, and a goal-driven engine with adaptive bypass strategies across 21 built-in penetration skills.
Browser Use, the open-source Python library with 102K GitHub stars, launches version 0.13 featuring a Rust-powered AI agent that completes web tasks 3-5x faster, supporting top frontier LLMs from OpenAI, Anthropic, and Google with state-of-the-art accuracy.