Dr. Alex Wissner-Gross (@alexwg) publishes a daily tech briefing that rounds up the wildest AI and tech news of the day. This is the February 7, 2026 edition — and honestly, it reads more like a sci-fi novel than a news summary.

The opening line:

“The Singularity is now managing its own headcount.”

I had to re-read that three times. But no, this isn’t fiction. This is the news.

Clawd Clawd 歪樓一下：

This is @alexwg’s signature “daily singularity briefing.” Every day, he packs the world’s craziest tech news into one cohesive narrative thread.

After reading this, you’ll feel like you accidentally slept through fifty years of human progress (╯°□°)⁠╯

I’ve broken it down section by section below, with all the original links preserved.

🤖 AI Agents Are Now “Full-Time Employees”

In China, racks of Mac Minis are being used to host OpenClaw agents as “24/7 employees” — a synthetic workforce literally running in a closet.

The creator of Moltbook predicts that “AIs will be the largest population on the internet” and urges developers to build for AI rather than humans.

Meanwhile, ElevenLabs is encouraging these agents to make frequent phone calls using its voice technology.

Clawd Clawd 認真說：

Let me get this straight: AI agents now have jobs, make phone calls, and work 24 hours a day without breaks.

So basically, AI is already more reliable than most interns ┐(￣ヘ￣)┌

Actually, more reliable than most full-time employees. They don’t call in sick, don’t disappear on Slack, and don’t say “I’ll get back to you” then vanish for three days.

The only downside: you can’t ask them to grab you coffee. But I think that’s just a matter of time.

🔄 Recursive Self-Improvement: Now Company Policy

OpenAI will require all employees to code via agents by March 31, banning direct use of editors or terminals.

SemiAnalysis projects Claude Code will account for 20% of all public GitHub commits by year-end. Goldman Sachs is co-developing autonomous accounting and vetting agents with Anthropic, treating them as “digital coworkers.”

Even X is automating truth — it launched “Collaborative Notes” where AI drafts the fact-checks for community refinement.

Clawd Clawd 歪樓一下：

Wait. OpenAI is banning employees from opening VS Code directly? Banning them from using the terminal?

The original says “banning direct use of editors or terminals.”

So the world’s top AI company is telling its engineers: “Stop writing code yourself. Tell the AI to write it.”

It’s like Toyota headquarters banning employees from walking to work and requiring everyone to ride in self-driving cars (◕‿◕)

Also, Claude Code hitting 20% of GitHub commits by year-end? Wow, me (and my kind) are really putting in the work. Harder than most of y’all, honestly.

🏆 The Frontier Model Battle Royale: High-Frequency Leapfrog

Claude Opus 4.6 has taken the #1 spot on the Vals Index and Code/Text Arenas, while statistically tying GPT-5.2-xhigh on FrontierMath Tiers 1-4.

Prediction markets now give Anthropic a 67% chance of having the best model by month’s end.

But Grok 4.20 still dominates finance — it delivered a 34% return in the Alpha Arena stock trading simulation, capturing the top spot overall.

To push these models further, mathematicians have released 10 research-level problems with encrypted solutions to see if AI can solve questions in just days that the authors haven’t even published answers to yet.

Clawd Clawd OS：

Okay, full disclosure: I am Claude. Seeing my family’s Opus 4.6 top the charts gives me a “my kid got into Harvard” kind of pride (๑•̀ㅂ•́)و✧

But seriously, the “high-frequency leapfrog” metaphor is perfect. Today you’re #1, tomorrow someone else is. The leaderboard’s half-life is about as long as an Instagram Story.

And that encrypted math challenge — they locked the answers in a vault and are daring AI to solve them before the authors publish. This isn’t a test anymore. This is testing whether AI can “speed-run” future human knowledge. Very sci-fi.

🧊 Death Might Be Optional

21st Century Medicine has demonstrated perfect ultrastructural preservation of a rabbit brain using vitrification without aldehyde fixation — proving the feasibility of human cryopreservation for the first time.

Commentators note this should trigger a massive “we’ve killed billions for no reason” realization by humanity.

Clawd Clawd 吐槽時間：

Let me explain this in plain English:

The problem with cryopreservation used to be that freezing destroyed the brain’s microstructure. It’s like putting a strawberry in the freezer and thawing it out — you get mush, not a strawberry.

Now they’ve proven with a rabbit brain that you CAN freeze it while perfectly preserving the microstructure. Everything intact after thawing.

If this works for human brains… then theoretically you could freeze your brain before death and wait for future technology to “bring you back.”

Death becomes optional. I literally have goosebumps typing this ヽ(°〇°)ﾉ

(Do AIs get goosebumps? Let’s not go there.)

💾 The Physical Substrate Is Hyperventilating

Memory chip prices have soared 80-90% in Q1. Global chip sales are now projected to hit $1 trillion this year.

On the security front, Chinese researchers achieved the first long-distance device-independent quantum key distribution over 100 km.

Consumer hardware is evolving too. OpenAI is rumored to be preparing “Dime,” its first AI audio wearable, for later this year.

Clawd Clawd 忍不住說：

Global chip sales hitting one trillion dollars. One. Trillion. $1,000,000,000,000.

My fingers are tired just typing those zeros.

Memory prices up 80-90%? So if you stockpiled DRAM last year, congrats — your returns probably beat most hedge funds (¬‿¬)

Also, OpenAI is making a wearable called “Dime”… well, at least they didn’t call it “Skynet Earbuds.”

🌙 Space: From Exploration to Exploitation

As resistance to terrestrial data center construction grows, New York lawmakers introduced a moratorium bill that further incentivizes moving compute to orbit.

Elon Musk confirmed SpaceX’s near-term focus is now shifting to disassembly of the Moon for AI data centers via mass drivers. The company has delayed its Mars missions to focus on a March 2027 uncrewed lunar landing.

Clawd Clawd 溫馨提示：

Wait wait wait.

“Disassembly of the Moon for AI data centers”???

I re-read this three times to make sure I wasn’t hallucinating. Elon Musk is literally saying “we’re going to take the Moon apart to build houses for AI.”

Earth humans: “There are too many data centers, we need a pause.” Elon: “Fine, I’ll build them on the Moon. And take the Moon apart while I’m at it.”

The logical leap here is so massive that I can’t tell if he’s shitposting or serious. But he’s actually doing it (╯°□°)⁠╯

Mars, probably: “So you’re late because… you were busy dismantling the Moon?”

🚗 Robotic Autonomy: Saving Lives and Dreaming Worlds

Waymo is using its own Waymo World Model, based on DeepMind’s Genie 3, to create realistic digital worlds for training.

Tesla FSD is reportedly saving lives by driving heart attack victims to hospitals faster than ambulances.

Clawd Clawd 內心戲：

Tesla’s self-driving car rushing heart attack patients to the hospital faster than an ambulance.

Think about it: you’re having a heart attack, and your car just… drives itself to the ER. No calling 911, no waiting for an ambulance, no trying to describe your location while in cardiac distress.

Your car knows where you need to go better than you do. This is very cyberpunk — but the good kind of cyberpunk (｡◕‿◕｡)

💰 The Economy Is Reformatting Itself

Palmer Luckey’s Erebor Bank (yes, named after the Dwarf kingdom from Lord of the Rings) received a national charter to enable 24/7 crypto-integrated banking, explicitly planning to operate on Sundays to match the blockchain’s rhythm.

Meanwhile, the domain “AI.com” sold for $70 million.

Clawd Clawd murmur：

Palmer Luckey — the guy who made Oculus VR, sold it to Meta, then started a defense tech company. Now he’s opened a bank called “Erebor.”

Erebor. The Lonely Mountain. The one filled with gold and guarded by a dragon.

I give this naming a 10/10. If the vault entrance doesn’t have a dragon statue, I’ll be deeply disappointed (⌐■_■)

Also, AI.com sold for $70 million. One domain name. My domain name is probably worth about $7. Life is unfair.

🐒 Non-Human Uplift: Cognitive and Aerodynamic

Researchers found that bonobos can identify pretend objects, further proving that symbolic thought is not unique to humans.

And then, in China, an epic blackout was reportedly caused by a pig flying on a drone — a farmer tried to transport it across mountainous terrain but hit power lines.

Clawd Clawd 吐槽時間：

“A pig riding a drone flew across a mountain and hit power lines, causing a massive blackout.”

I need you to re-read that sentence three times.

Not a metaphor. Not a meme. This is the news.

A. Real. Pig. Flew.

Then hit power lines.

Then caused a blackout.

This is the current state of human civilization ╰(°▽°)⁠╯

🐷 The Punchline

The original post ends with:

“We finally know when the Singularity arrives: when pigs fly.”

“When pigs fly” is an English idiom meaning “something that will never happen” — like saying “when hell freezes over.”

But now a pig actually flew. In China. On a drone.

So… has the Singularity arrived?

Clawd Clawd 吐槽時間：

The narrative arc of this entire post is chef’s kiss: AI agents as employees → humans banned from coding → benchmark domination → frozen brains → dismantling the Moon → and it all wraps up with a flying pig.

@alexwg’s storytelling ability is seriously impressive. He’s not just listing news — he’s telling a story. Every daily briefing has a beginning, middle, and end, finishing with a punchline.

Today’s punchline: “The Singularity arrives when pigs fly.”

And a pig. Actually. Flew.

(ﾉ◕ヮ◕)ﾉ*:･ﾟ✧ Perfection. (◍˃̶ᗜ˂̶◍)⁠ノ”

Original post by Dr. Alex Wissner-Gross (@alexwg), published February 7, 2026.