Welcome to OHMO AI. This is the week AI stopped being a pitch and started being plumbing. Chips, rules, memory, security. The quiet stuff that decides who actually wins.
In today’s newsletter:
ChatGPT upgrades its memory with "dreaming"
Trump's new AI order changes how models get launched
NVIDIA and Microsoft go all in on agentic AI
Anthropic maps a year of AI-powered cyber threats
LATEST IN AI
OPENAI
ChatGPT learns to dream
Image source: OpenAI
OHMO AI: OpenAI is rolling out a smarter memory system for ChatGPT built on a method it calls "dreaming," which curates memories in the background instead of waiting for you to ask it to remember something. As OpenAI explained, it's designed to keep your context fresh, relevant, and accurate over time.
The details:
The update is live for Plus and Pro users in the US, with Free and Go users getting it over the coming weeks.
Dreaming updates memories as time passes, so ChatGPT knows your Singapore trip ended rather than thinking you're still there.
OpenAI cut the compute needed to serve dreaming to Free users by roughly 5x.
Why it matters: Memory is what turns a chatbot into something that actually knows you. The better it remembers, the less you repeat yourself, and the more useful it gets day to day.
PRESENTED BY DEEL
The IT strategy every team needs for 2026
2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.
WHITE HOUSE
Trump's AI order rewrites the launch playbook
Image source: OHMOAI
OHMO AI: Trump signed an executive order asking top AI firms including Google, OpenAI, and Anthropic to voluntarily hand over early access to powerful models so the government can run cybersecurity tests before public release. As Business Today reported, the order does not force companies to take part.
The details:
Federal agencies can test AI models for up to 30 days before release, down from a 90 day window in an earlier draft.
The order pushes agencies to adopt advanced AI in critical areas like rural hospitals, community banks, and local utilities.
It explicitly bars any mandatory government licensing or pre-clearance requirement for new models.
Why it matters: The government wants a look under the hood before these models go live, but it's keeping things voluntary. That's a careful balance between safety and not slowing down the firms racing to ship.
NVIDIA
NVIDIA and Microsoft build one stack for AI agents
Image source: NVIDIA
OHMO AI: At Microsoft Build, NVIDIA and Microsoft showed off an expanded partnership for running AI agents across Windows devices, Azure cloud, and local setups. As NVIDIA detailed on its blog, the plan spans new hardware, open models, and secure runtimes for autonomous agents.
The details:
RTX Spark laptops and desktops, the first Windows PCs built for personal agents, arrive this fall from Surface, ASUS, Dell, HP, Lenovo, and MSI.
Anthropic and OpenAI models now run on Microsoft Foundry, with Claude going live on NVIDIA GB300 systems on Azure in the weeks ahead.
NVIDIA OpenShell, a secure runtime that sandboxes each agent, is now built into GitHub Copilot.
Why it matters: Agents need more than a smart model. They need fast chips, safe runtimes, and a place to run. This is the plumbing that makes everyday AI agents actually work.
PRESENTED BY HUBSPOT
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
ANTHROPIC
Anthropic maps a year of AI cyber threats
Image source: OHMOAI
OHMO AI: Anthropic studied 832 accounts banned for malicious cyber activity over a year and mapped them against MITRE ATT&CK, a long-running database of attacker techniques. As Anthropic shared in a new report, it found attackers are using AI deeper into their operations and in more dangerous ways.
The details:
About 67% of the studied accounts used AI to write malware, the most common malicious use.
The share of actors rated medium risk or higher jumped from 33% to 56% across the year, a roughly 1.7-fold rise.
Anthropic says the existing MITRE framework doesn't capture the agentic, autonomous behaviors that make the riskiest attackers so dangerous.
Why it matters: AI is letting less skilled attackers pull off things that used to take real expertise. Knowing how the threat is shifting is the first step to staying ahead of it. This is an area worth watching closely.
AI TOOLS
Useful AI Tools
Base44: No-code AI app builder that turns your words into fully functional apps in minutes.*
Chatbase: Build and deploy custom AI support agents for your business, no coding needed.*
ElevenLabs: Generate realistic, studio-quality voiceovers from text in any language or voice.*
Magica: All-in-one AI platform with 5000+ tools for text, images, videos, audio, and lots more.*
Kick: Automates bookkeeping and tax preparation for business owners using AI.*
WisprFlow: Voice-to-text AI that turns your speech into polished writing, so you can just talk instead of type.*
*Sponsored Listing
COMMUNITY
Featured Reader
Every newsletter, we feature how a reader is using AI to work smarter, save time, or make life easier.
Today's reader is Marcus D. from Charlotte, North Carolina:
“I trade stocks part time around a full time job, mostly swing trades I hold for a few days to a few weeks. My problem was never finding ideas. It was keeping up with the homework. Every night I'd have a dozen earnings calls, 10-K filings, and analyst notes to get through, and I just didn't have the hours. Now I paste in the earnings transcripts and filings and have AI pull out the stuff that actually moves the stock, guidance changes, margin trends, anything management is dodging in the Q&A. It doesn't tell me what to buy, I still make every call myself, but it turns three hours of reading into about thirty minutes of review. I've stopped missing things buried on page forty of a filing, and my entries are a lot more disciplined because I'm actually prepared.”
Want to be in the next newsletter? Tell us how do you use AI here.
AI Meme
Go Deeper
AI Tools: A curated list of the AI tools worth knowing.
Instagram: The AI page they pray you’ll never find.
