Agentic AI News

China’s central government has moved decisively against OpenClaw AI, restricting its use at state-run banks and government agencies — even as thirteen of the country’s biggest technology companies are simultaneously racing to integrate or fork the platform into their own products. The split tells a story about how authoritarian states navigate powerful foreign AI: ban it at the top, absorb it at the bottom. What Beijing Actually Said According to reporting confirmed by Bloomberg, Reuters, and multiple regional outlets, Chinese authorities have instructed state-run enterprises and government agencies to remove OpenClaw AI apps from office computers. The stated rationale is cybersecurity and data-leak risk — the same concern Beijing has raised about prior foreign software platforms, from Windows to Slack. ...

Anthropic has built its brand on safety. Claude is consistently positioned as the thoughtful, cautious model — the one that pushes back on dangerous requests, that thinks about consequences, that errs on the side of care. So the DryRun Security research published today will raise some eyebrows: when used as an agentic coding agent building real applications, Claude produces the highest number of unresolved high-severity security flaws among the leading AI coding agents tested. ...

JetBrains launched Air into public preview today — and it’s not just another AI-enhanced code editor. It’s a fundamental rethink of what a development environment is when AI agents do the actual coding. The tagline from JetBrains captures the shift cleanly: traditional IDEs add tools to the code editor. Air builds tools around the agent. What Air Actually Is Air is JetBrains’ answer to the question: “What does an IDE look like if the primary user is an AI agent, not a human?” The answer involves several distinct design choices that set it apart from tools like Cursor, GitHub Copilot, or even the JetBrains AI plugin for IntelliJ IDEA. ...

Perplexity just announced something that sounds deceptively simple: an AI agent that lives on your Mac mini, runs 24 hours a day, and has access to your local files and applications. They’re calling it Personal Computer, and they unveiled it at their Ask 2026 developer conference. Waitlists opened immediately. This is more consequential than another AI announcement. Perplexity is making a direct architectural bet against the cloud-first model that dominates agentic AI today. ...

The agentic IDE war is moving fast. JetBrains launched Air in public preview just yesterday, positioning itself as the agent-first development environment. Today, Zed — the performance-obsessed open-source editor — ships v0.227.1 with a feature that directly answers the multi-agent challenge: native parallel subagents. What’s New in v0.227.1 The headline feature is a new spawn_agent tool built into Zed’s AI assistant. When the AI determines a task would benefit from parallel execution, it can now spawn and run subagents in parallel — each handling a discrete portion of the work in its own context window — and then coordinate the results. ...

GitHub has moved Copilot CLI to general availability for all Copilot subscribers, bringing full agent mode to the terminal. The GA release, announced via the official GitHub changelog, adds the capability to plan, write, debug, and review code without leaving the shell — along with custom agent definitions and sub-agent support for JetBrains IDEs. For developers who live in the terminal, this is the Copilot integration they’ve been waiting for. The browser and IDE-based Copilot experiences have matured significantly, but the command line remained a second-class citizen until today. ...

Every year the threat intelligence industry produces a report that crystallizes what defenders already suspected but couldn’t quite prove. Flashpoint’s 2026 Global Threat Intelligence Report (GTIR) is this year’s version — and its central claim is blunt: agentic AI has crossed from criminal curiosity to deployed offensive infrastructure. This isn’t speculation. It’s sourced from Flashpoint’s Primary Source Collection (PSC), which monitors criminal forums, dark web markets, and adversarial communication channels in near-real-time. The signal they’re seeing is a rapid acceleration of AI-related discussions that started as curiosity and has hardened into active capability development. ...

Anthropic is doing two things at once: building the most sophisticated AI policy apparatus in the industry, and fighting for its survival against a federal government that has designated it a supply-chain risk. On Wednesday, the company announced the Anthropic Institute — a new internal think tank combining three existing research teams — while simultaneously disclosing that the White House is preparing another executive order that could threaten hundreds of millions in 2026 revenue. ...

Two major enterprise AI reports landed on the same day — and they’re telling the same uncomfortable story: organizations are racing to deploy AI agents, but the majority can’t tell you whether those agents are actually delivering value. ModelOp’s 2026 AI Governance Benchmark Report and Gartner’s Data & Analytics Summit Day 2 both focused heavily on agentic AI adoption in the enterprise — and both landed on the same core finding: the tools are proliferating, but the measurement and governance infrastructure hasn’t kept pace. ...

Yann LeCun — Turing Award winner, longtime Meta chief AI scientist, and one of the most vocal critics of the current large-language-model paradigm — has officially left the building. His new venture, AMI Labs (Advanced Machine Intelligence), just closed a $1.03 billion seed round at a $3.5 billion pre-money valuation, making it Europe’s largest-ever seed raise in tech. The backers include Nvidia, Temasek, and Jeff Bezos. The thesis: build AI that understands physical reality, not just language. ...