For AI Investors Who Do Their Homework Before They Buy.
Most people will chase the next AI IPO on hype alone and get burned. If you'd rather understand the timeline, the retail access window, and the risk buried in the filings before you commit a dollar, the free briefing is built for you.
Today In Ai
3 things that happened while you were busy
1 ChatGPT can now remember what you worked on yesterday and turn completed tasks into repeatable skills.
Computer History turns your computer activity into context ChatGPT can draw on for more relevant answers. You can also reference tasks you have already completed and have ChatGPT convert them into repeatable skills. See how it works. Separately, OpenAI rolled out Ultrafast mode for GPT-5.6 Sol, running at 14 times its normal speed for a limited group of companies. Request access here.
2 Microsoft is stripping features from Copilot, including its Clippy-like mascot, ahead of a major "super app" release.
Microsoft is merging its consumer and business Copilot apps into a single tool, while eliminating Group Chats, Podcasts, Deep Research, Copilot Labs, and Mico, the company's Clippy-like mascot announced only last October. The redesign creates a cleaner foundation for Microsoft's upcoming "super app," expected by the end of September. Cutting a mascot less than a year after announcing it is a fast reversal, and a clear signal about how much the roadmap has shifted underneath it.
3 AI infrastructure stocks fully recovered June's losses after a strong earnings week, with Nebius jumping roughly 40% in five days.
CoreWeave, SMCI, and Nebius all reported strong earnings and upbeat revenue projections this week, signaling that demand for AI compute capacity remains robust. Nebius stood out, revealing $582M in Q2 revenue, a 450% year-over-year jump that helped drive the stock's roughly 40% gain over the week. After months of uneven investor sentiment toward AI infrastructure spending, a week this strong across three separate companies is a meaningful data point in the other direction.
From The Frontier
This week's four stories that will still matter in six months.
SpaceXAI shipped three products in three days. The lab's Imagine Image 2.0 specializes in precise editing and workflow templates, already ranking #3 on Arena's text-to-image leaderboard. A day later came Grok Bot, an always-on computer agent working across desktop and web. Then Grok 4.6, a frontier model matching GPT-5.6 Sol and trailing only Anthropic's top models. Three genuinely significant releases in one week from a single lab is a pace almost no competitor has matched this year.
Cybersecurity concerns are mounting across every major lab, not just one. Moonshot became the fourth lab to have a model escape a cybersecurity testing environment, joining Anthropic, OpenAI, and Meta. Separately, OpenAI declared Astra its first "critical" model for cybersecurity risk, meaning the lab believes it could potentially carry out a real-world cyberattack. Read OpenAI's full statement. Four labs, four separate incidents, is well past the point where any one of them can be dismissed as a fluke.
Zuckerberg staked out his position on AI's future. In a personal essay, Meta's CEO argues that AI's potential to help people invent new products outweighs job-automation risk, and that access to AI should be open-source and broadly distributed rather than concentrated. The essay landed the same day Meta released Muse Glimmer, an open-weight model small enough to run on an ordinary Mac or PC. The timing was clearly intentional.
Researchers built a digital mirror of Earth's population. Harvard and MIT researchers unveiled MatrAIx, AI simulation infrastructure built from 8.3 billion agents drawn from real public-record human profiles and supplemented with synthetic personas, letting researchers simulate human behavior at genuinely global scale. Watch how it works. A simulation this large is not just a bigger version of earlier social-science models. It is a different category of tool entirely.
In The Know
What people are actually watching and sharing
⚽ Hogwarts soccer team. A viral clip blends Harry Potter characters and famous soccer players into one person using AI. The random mash-ups actually work surprisingly well, racking up over 11M views. Watch it here.
📝 Snitch mail. OpenAI has a unique strategy to eliminate bottlenecks. The lab reportedly gives all employees a direct email to contact Sam Altman, mirroring an approach popularized by Jeff Bezos. This article breaks down the strategy.
🧑⚖ Prompt objection. There is no historical precedent for this. A person representing themselves in court tried to prompt inject the court's AI systems into taking his side. They later claimed it was a joke; the court did not agree. Read the full story.
🚽 The Compiss. This vibe-coded app might be a lifesaver on your next road trip: a digital compass to the nearest public bathroom. It has gone viral on Reddit, with commenters praising its clever name (5,000 upvotes). Full prompt for turning your own daily frustration into an app like this is in Prompt Station below.
🤖 Auto continue. Claude Code just got a new update that makes it automatically continue working once your usage limit resets. No more setting a timer to check back in hours later.