OpenAI released GPT-5.5-Cyber to verified defenders, calling it its strongest model yet for finding and patching software vulnerabilities. The same AI muscle that helps attackers is now being pointed the other way — carefully.
Gartner says AI agent software spending will more than double to $206.5 billion this year, and a run of acquisitions shows what vendors are really paying for: the plumbing that lets agents act, not just advise.
AWS put a FinOps Agent into public preview that investigates your cloud bill, files the ticket, and pings Slack — but won’t resize or delete a single resource. That restraint is the whole point, and it’s a lesson for anyone deploying agents.
Salesforce is buying Fin, the AI support agent formerly known as Intercom, for about $3.6 billion. Fin already resolves 76% of support volume end to end. Here’s what the deal signals for SMBs and enterprise IT.
AI-generated phishing is the defining security threat of 2026, and the old spam filter isn’t going to save you. Here are the five AI security tools I’d actually put my money behind this year — what they cost, what they do, and who each one is for.
Google shipped Gemini 2.5 Pro with Deep Think on June 22, and the benchmarks put it at the front of the pack. Here’s what actually matters if you use AI tools every day — and what doesn’t.
A cascade of agent-governance frameworks landed this week as the EU AI Act’s high-risk clock ticks toward August 2. The most uncomfortable data point: the most careful companies are pulling their agents the fastest.
Gartner pegs purpose-built AI agent software at $206.5 billion in 2026 — a 139% jump in one year. Meanwhile the agent-building tools are landing fast. Here’s what’s real and what to do with it.
I run Claude Cowork at home and Microsoft Copilot Cowork at work, every day. Updated June 2026 for Copilot Cowork’s general-availability launch and its new usage-based pricing. Here is where each one actually wins, and the friction nobody else writes about.
FIRST just raised its 2026 vulnerability forecast to roughly 66,000 CVEs, and the reason is blunt: AI agents are hunting software flaws on their own. Here is what that means whether you run a one-person shop or an enterprise security team.