Posts

Microsoft’s enterprise AI story just got a lot more interesting — and more nuanced — than a simple “Claude checks GPT’s homework” framing suggests. The short version: Microsoft 365 Copilot’s agent governance capabilities are now generally available, with Claude integrated as a Microsoft-approved AI subprocessor operating alongside GPT-4o in enterprise workflows. Screen agents are live for eligible enterprise customers. And a new M365 E7 tier bundles all of it at $99 per user per month. ...

When Tibo Sottiaux, OpenAI’s head of Codex product, posted on X on May 31, 2026 to announce a usage milestone, the context was telling. He was replying to developer Siqi Chen, who had written about switching away from Codex for a few days and being struck by how much better it had gotten. “Five million users would agree,” Sottiaux replied. “Resetting the limits tomorrow morning to celebrate.” The number itself — whether it’s exactly 5 million or approaching it, with some independent analyst estimates closer to 4 million weekly active users globally — is secondary to what the trajectory represents. ...

On May 26, 2026, Gartner published a press release that should be required reading for every enterprise CTO who’s been deploying — or planning to deploy — autonomous AI agents. The headline finding is stark: applying uniform governance to AI agents will cause widespread enterprise failure. Not might cause. Will cause. Gartner’s specific prediction: by 2027, 40% of enterprises will demote or decommission autonomous AI agents due to governance gaps that surface after production incidents. Not because the agents didn’t work. Because the governance structure failed to match the risk profile of what was deployed. ...

If you’re reading this on May 31, 2026, you have roughly 24 hours to understand a billing change that affects every GitHub Copilot user. Starting June 1, 2026, GitHub Copilot drops the Premium Request Units (PRU) model and switches to AI Credits — a usage-based system tied directly to token consumption. Here’s what you need to know before the clock resets. The Core Change: From PRUs to AI Credits The old system counted “premium requests” — a blunt instrument that treated every chat message or agent interaction as a discrete unit regardless of how much compute it actually consumed. Starting June 1, the new system counts tokens. ...

Something significant just happened in the AI infrastructure space, and the significance is precisely how unremarkable each individual announcement looked. Within a six-week window in spring 2026, three of the biggest players in AI — Google, Anthropic, and AWS — all shipped managed runtimes for AI agents. Independently. Without coordinating. And yet they all arrived at essentially the same feature set. That’s not a coincidence. That’s a market signal. The Three Runtimes Google Agent Executor (AX) — May 20, 2026 Google open-sourced Agent Executor (GitHub: google/ax) on May 20 as a distributed runtime standard for reliable AI agent execution in production environments. The key capabilities: ...

When historians look back at the moment AI agents crossed into mainstream consumer finance, May 27, 2026, is going to be the date they circle. Robinhood — the platform that democratized stock trading for millions of retail investors — just opened its doors to AI agents. Not its own AI agents. Your AI agents. What Robinhood Actually Launched Let’s be precise here, because the framing matters: Robinhood is not shipping its own AI trading bot. What it’s doing is more significant. It’s building the rails that let you connect your own AI agent — Claude, GPT-4o, Codex, or anything else — directly to a live Robinhood brokerage account via the Model Context Protocol (MCP). ...

On May 30, 2026, Simon Willison posted what might be the most practically useful thing written about AI agent safety this year: a close read of Anthropic’s engineering post on how they contain Claude across their products. If you build agentic AI systems, this is required reading — and here’s why. The Gap Willison Identified Willison opens with a complaint that resonates with anyone who’s evaluated sandboxing products: they’re rarely thoroughly documented. You’re handed marketing claims about “secure execution environments” with little detail about what that means in practice. Without documentation, you can’t reason about what you’re trusting or where the gaps might be. ...

🔒 Security advisory: AG2 v0.13.2 patches GHSA-9fvw-gr53-m7fw, a code injection vulnerability affecting ContextExpression in prior versions. If you’re running AG2 in production, upgrade immediately. AG2 v0.13.2: Evaluations, New Clients, and a Critical Security Patch AutoGen — now officially maintained as AG2 — shipped v0.13.2 on May 29, 2026, with three notable additions landing in one release: a beta evaluation framework for grading agent outputs, two new V2 LLM clients expanding model access, and a critical security fix that should make upgrading non-optional for anyone running AG2 in production. ...

⚠️ Transparency note: This story is based on leaked materials and independent journalist reporting. Microsoft has not officially confirmed these products, features, or timelines. Treat all details as unverified until a formal Microsoft announcement. Microsoft’s Copilot Super App Has a Proactive Secret: Meet Scout Microsoft isn’t just iterating on Copilot — it’s rebuilding the whole thing. Leaked screenshots published by journalist Alex Heath at Sources.news, combined with independent reporting from Fortune’s Sebastian Herrera, reveal Microsoft is developing a unified Copilot super app that consolidates its scattered AI tools into a single surface. And tucked inside that app is a previously unreported feature that will sound very familiar to anyone in the agentic AI space: a proactive, always-on agent called Scout. ...

The AI Agent Security Market Just Got Its First Major Consolidation At 8:00 AM Pacific on May 29, 2026, Palo Alto Networks announced it had closed the acquisition of Portkey — the widely-used AI gateway platform — for approximately $700 million. The deal, announced on April 30 and now complete, represents the first major consolidation in what’s becoming one of the most contested spaces in enterprise software: AI agent security and governance. ...