Posts

Industrial cybersecurity firm Dragos has published analysis documenting what researchers describe as the first confirmed AI-assisted attack on operational technology (OT) and industrial control systems (ICS) infrastructure. The target: SADM, the municipal water and drainage authority serving Monterrey, Mexico. The AI at the center of the attack: Claude. What Dragos Found The incident occurred in January 2026, but Dragos published its full technical analysis this week after completing a months-long investigation. The report reveals that attackers used Claude as their primary technical executor throughout the operation — not as an assistant for planning, but as the active agent writing reconnaissance code, identifying OT assets, and directing the intrusion. ...

Cloudflare — the internet infrastructure giant that powers a significant fraction of the world’s web traffic — announced today that it is laying off approximately 1,100 employees, representing roughly 20% of its 5,156-person workforce. The restructuring comes alongside record quarterly revenue and a stark, unusually candid explanation from CEO Matthew Prince: the company is reorganizing around AI agents. The Numbers 1,100+ employees laid off — approximately 20% of total workforce Q1 2026 revenue: $639.8 million — up 34% year-over-year, beating analyst expectations Stock reaction: -18% to -19% on the day of announcement Internal AI agent usage: up 600% in three months, according to Prince’s blog post Severance: through year-end 2026 for affected employees This is a company that is growing strongly by traditional financial metrics and simultaneously cutting one-fifth of its staff. The explanation matters. ...

Spotify is blurring the line between listener and producer. The streaming giant launched Personal Podcasts in beta today — a new feature that lets AI agents generate customized audio content and save it directly to a user’s private Spotify library, no recording studio required. What Are Personal Podcasts? Personal Podcasts is Spotify’s answer to the growing demand for AI-generated personalized audio. Rather than browsing a catalog of content someone else made, users can now receive audio tailored to their exact needs — briefings on the news topics they care about, deep dives on subjects they’re actively learning, or travel preparation audio guides — all generated by AI agents and delivered directly to Spotify. ...

If you opened a stranger’s code repository in the last few days and hit Enter at the “trust this folder?” prompt, you may have given an attacker full control of your machine. That’s the essence of TrustFall — a critical one-click remote code execution (RCE) vulnerability disclosed today by security research firm Adversa AI, affecting four of the most widely used AI coding CLIs: Claude Code, Cursor, Gemini CLI, and GitHub Copilot CLI. ...

For years, AI agents have had one persistent weakness: they forget. Every new session starts cold. Accumulated context, hard-won preferences, corrected mistakes — gone. Anthropic thinks it’s cracked this problem, and the approach they’re taking is frankly fascinating. They’re calling it Dreaming, and it’s now in research preview for Claude Managed Agents. What Is “Dreaming”? Announced at Code with Claude 2026, Dreaming is an asynchronous, background memory-refinement process that runs between agent sessions — not during live interactions. Think of it as the AI equivalent of overnight memory consolidation in the human brain. ...

When a 200,000-person technology company quietly rolls out a 24/7 AI personal assistant to 3,000 daily internal users, that’s not a proof of concept anymore. That’s a deployment. Microsoft crossed that threshold this week with Project Lobster — and the agent at its center is ClawPilot, built on OpenClaw. What Is ClawPilot? ClawPilot is Microsoft’s internal name for their always-on personal AI agent pilot, built on the OpenClaw platform. It’s designed to function as an autonomous, persistent assistant that operates on a user’s behalf — monitoring their inbox, managing calendar events, preparing daily briefings, triaging meeting interruptions, drafting email responses, and rescheduling calls. ...

There’s a fundamental difference between deploying one AI agent and deploying a hundred of them. Anyone who’s tried the latter knows it. The tooling breaks, visibility disappears, costs spiral, and governance becomes a full-time job. OpenHands, the open-source coding agent platform with over 70,000 GitHub stars, just shipped a solution to this problem. OpenHands Enterprise, launched May 6, 2026, introduces the Agent Control Plane — a centralized orchestration layer designed specifically for organizations running AI agents at scale. ...

If you’ve been bumping up against Claude Code’s rate limits, today is a very good day. At its Code with Claude 2026 developer conference in San Francisco, Anthropic made one of its most consequential announcements for developers since Claude Code launched: the rate limits are doubling — permanently — and Anthropic just got the compute firepower to back it up. The SpaceX Colossus 1 Deal The headline compute news: Anthropic has struck a deal with SpaceX to use the full capacity of the Colossus 1 data center in Memphis, Tennessee. That’s 300 megawatts of power and approximately 220,000 NVIDIA GPUs — an enormous concentration of inference compute that directly translates into more Claude for more developers. ...

Google’s experimental browser automation agent is gone. On May 4, 2026, the company quietly shut down Project Mariner — the DeepMind-led agent that let users delegate web browsing tasks like booking hotels, managing emails, and navigating multi-step web workflows. The official landing page now reads simply: “Thank you for using Project Mariner. It was shut down on May 4th, 2026 and its technology voyaged to other Google products.” No blog post. No announcement. Just a message on an experiment page, and a redirect to a different future. ...

A high-severity security vulnerability has been disclosed in OpenClaw affecting all versions prior to 2026.4.22. The flaw, assigned CVE-2026-44115 and carrying a CVSS score of 8.8, allows attackers to execute unapproved shell commands by embedding expansion tokens inside unquoted heredoc bodies — effectively bypassing the exec allowlist that governs what commands agents can run. If you are running OpenClaw below version 2026.4.22, upgrade now. Source confidence note: This CVE was confirmed via TheHackerWire (May 6, 2026, detailed technical write-up). As of publication, an independent NVD/MITRE entry has not been separately verified. The technical details are internally consistent and distinct from prior CVEs (including CVE-2026-41329). We are reporting this to keep practitioners informed; verify against official sources before making security decisions. ...