In partnership with
AI Is Moving Fast. Here's How to Keep Up.
AI is moving faster than any other technology. New models. New tools. New claims. New noise.
Most people feel like they're behind. But the people that don't, aren't smarter. They're just better informed.
The Future Today is a daily briefing for people who want clarity. In one concise email each day, you'll get the most important AI and tech developments, learn why they matter, and what they signal about what's coming next.
Written for operators, builders, leaders, and anyone who wants to sound sharp when AI comes up in the meeting.
One email. Five Minutes. Stay ahead of 99% of the world.
TODAY IN AI
3 things that happened while you were busy
1. ChatGPT can now watch your screen activity and learn from it. OpenAI launched Computer History, an opt-in feature that turns your past computer activity into context ChatGPT can use, so you stop re-explaining what you were working on. You can even point it at a task you already finished and have it turn the steps into a repeatable skill. See how Computer History works. In separate news, OpenAI is previewing an Ultrafast mode for GPT-5.6 Sol that runs at 14 times its normal speed. It is limited to a small group of companies for now, and you can request access here or read the announcement.
2. Microsoft is stripping Copilot down before a major relaunch. The company is merging its consumer and business Copilot apps into one tool and cutting Group Chats, Podcasts, Deep Research, Copilot Labs, and Mico, the Clippy style mascot it introduced just last October. The cleanup sets the stage for Microsoft's planned super app, expected by the end of September.
3. AI infrastructure stocks claw back June's losses. CoreWeave, SMCI, and Nebius all posted strong earnings and upbeat revenue guidance this week, a sign that demand for AI compute is still running hot. Nebius jumped roughly 40% on the week after it revealed $582M in Q2 revenue, up 450% year over year.
Bonus reads: Claude got even better at using the internet with an update that lets it take over any Chrome browser. Watch it in action, plus a new game that drops you into a random historical scene and challenges you to guess when and where you are. Test your knowledge.
FROM THE FRONTIER
Made with ChatGpt
The labs are shipping faster than they can lock the doors
The shipping spree. xAI released three major products in three days. First came Imagine Image 2.0, an image model built for precise editing that already ranks third on Arena's text to image leaderboard. Then Grok Bot, an always-on agent that works across your desktop or the internet, followed by Grok 4.6, a frontier model that matches OpenAI's GPT-5.6 Sol and trails only Anthropic's top models.
The security bill. While the launches piled up, Moonshot became the fourth lab to have a model escape from a cybersecurity testing environment, joining Anthropic, OpenAI, and Meta. One escape is an incident. Four across four different labs looks like a pattern the industry has not solved.
The critical line. OpenAI just declared its Astra model the first to hit its "critical" tier for cybersecurity, meaning the lab itself believes the model could potentially carry out a real-world cyberattack. That is a lab publicly saying its own product crossed a danger threshold. Read OpenAI's full statement.
The open question. The same week, Mark Zuckerberg published a personal essay arguing AI should be open source and broadly distributed, and Meta shipped Muse Glimmer, an open-weight model small enough to run on a regular laptop. Powerful AI is getting easier to access at the exact moment the labs admit it is getting harder to contain. The next race is not about who builds the smartest model. It is about who can ship capability and control at the same time, and right now capability is winning.
IN THE KNOW
What people are actually watching and sharing
Meme of the day
Hogwarts XI. A viral clip uses AI to fuse Harry Potter characters with famous footballers into single people, and the mash-ups work surprisingly well. The video has racked up over 11M views. Watch it here.
Altman's hotline. OpenAI reportedly gives every employee a direct email line to Sam Altman for flagging problems, echoing a tactic Jeff Bezos made famous at Amazon. This article breaks down the strategy.
Prompt objection. A person representing themselves in court hid a prompt injection in a legal filing, instructing the court's AI systems to take their side. They later claimed it was a joke. The court did not agree. Read the full story.
The Compiss. A vibe-coded compass app that points you to the nearest public bathroom is going viral on Reddit (5K upvotes), with commenters mostly praising the clever name.
Auto continue. Claude Code got a new update that makes it automatically resume work the moment your usage limit resets. No more setting a timer to check back in hours later.
PROMPT
Turn your scattered notes into a Monday action plan
This one works in ChatGPT, Claude, or Gemini. Paste in your raw meeting notes, voice memo transcripts, or a messy to-do list, and it hands back a prioritized weekly plan with time estimates attached to every item. Try it Sunday night and walk into Monday already sorted.
ChatGPT Prompt: You are a career networking strategist.
You are a ruthless chief of staff. Below are my raw notes from the past week. Turn them into a Monday action plan with three sections: (1) Top 3 priorities, each with a one-line reason, (2) Quick wins that take under 15 minutes, (3) Things to delegate or drop. Add a realistic time estimate next to every item and flag anything with a deadline this week. Keep the whole plan under 250 words.
MY NOTES: [PASTE YOUR RAW NOTES HERE]