Posts

OpenAI launched Daybreak on May 11, 2026 — a cybersecurity initiative that deploys frontier AI models as active participants in the software security lifecycle. Powered by GPT-5.5 (released April 23, 2026) and a new agentic layer called Codex Security, Daybreak is positioned as OpenAI’s answer to the growing consensus that AI must be part of software defense, not just software development. The name is intentional: Daybreak means seeing risk before it becomes an exploit. ...

OpenAI’s Workspace Agents — Codex-powered autonomous agents that live inside ChatGPT Enterprise — reached full Enterprise support on May 7, 2026. First introduced in April 2026, they represent a significant shift in how enterprises think about AI in the workflow: not as a tool employees use, but as a participant that runs tasks independently. Workspace Agents operate in the cloud and continue working even when the employee who created them is offline. ...

At SAP Sapphire 2026, one of enterprise software’s largest annual events, SAP and Anthropic announced a partnership that puts Claude at the center of SAP’s agentic AI strategy. Claude will be embedded across the newly announced SAP Business AI Platform, serving as the primary reasoning and agentic engine powering Joule — SAP’s AI assistant — and a growing fleet of Joule agents. The scale here is significant: 200+ domain-specific automations spanning ERP, supply chain, HR, finance, and procurement. ...

A new study from Akeyless puts a number on a concern that security practitioners have been raising for months: 67% of enterprise IT and security leaders suspect their AI agents have already accessed unauthorized data. Not might. Already. The finding comes from Akeyless’s 2026 State of AI Agent Identity Security Report, which surveyed 400 IT and security leaders across the US and UK. The picture it paints of enterprise AI agent security is alarming — not because of sophisticated attacks, but because of something more mundane and more dangerous: over-privileged credentials handed to agents that were never designed to be contained. ...

For enterprises who’ve been wrestling with separate accounts, separate billing, and the complexity of managing two distinct identity systems, the news landed quietly but meaningfully: Anthropic’s full native Claude Platform is now generally available through your existing AWS account. No separate sign-up. No parallel billing pipeline. No additional identity management overhead. Claude, the way Anthropic ships it, running on the infrastructure you already operate. What “Native Claude Platform” Actually Means This isn’t Claude-on-Bedrock, and that distinction matters more than it might seem at first glance. ...

Update post: We first covered the Hologram infostealer and fake OpenClaw installer campaign on May 8. New reporting from GBHackers and cross-validation across Netskope, SOCPrime, Malwarebytes, Cisco, and the 1Password blog has surfaced additional technical details that expand the picture. This post covers what’s new. The Hologram infostealer campaign targeting OpenClaw users has more teeth than the initial reports indicated. Multiple independent security researchers have now documented the campaign in detail, and the scope — both in terms of target breadth and infrastructure — is more extensive than the first wave of coverage captured. ...

The agentic AI wave has been crashing through enterprise software, developer tooling, and cloud infrastructure for months. Today it hits something more fundamental: the physical networks that everyone — not just developers — depends on. Nokia announced the launch of agentic AI capabilities across its fixed network management portfolio this week, targeting the exact operational pain points that have made broadband network management expensive, slow to respond, and chronically dependent on human field intervention. ...

At Red Hat Summit 2026 in Atlanta, the enterprise Linux giant made the clearest statement yet about how it intends to compete in the agentic AI era: not by abandoning the infrastructure enterprises already run, but by making that infrastructure first-class for agentic workloads. The headline announcements — AI Skills Repositories and the Fedora Hummingbird Linux distribution — share a common philosophy. Enterprises shouldn’t need to rebuild their entire stack to run AI agents effectively. Red Hat wants to be the layer that makes the stack they already have work for the AI-native future. ...

One of the foundational constraints of autonomous AI agents has always been that they can’t actually pay for things. They can call free APIs and open MCP servers, but the moment a resource requires payment, the agent hits a wall and waits for human intervention. Amazon just removed that wall. Amazon Bedrock AgentCore Payments, launched May 7, enables AI agents running in the AgentCore managed runtime to autonomously access and pay for APIs, MCP servers, web content, and other agents — using stablecoin micropayments via the x402 HTTP payment protocol, built in partnership with Coinbase and Stripe. ...

There’s a well-known problem in long-running AI agents: over time, their memory stores turn into landfills. Duplicate entries pile up, contradictions from earlier sessions persist alongside newer correct values, and the useful signal gets buried under noise. Anthropic’s new Dreaming feature is designed to clean that up — automatically, asynchronously, and without any weight updates. Announced at the Code with Claude conference and now live in research preview, Dreaming lets Claude Managed Agents run asynchronous background jobs that process their own memory stores and past session transcripts, producing a reorganized, higher-signal memory output. It’s the closest thing yet to an AI agent that genuinely improves itself between sessions. ...