Welcome, humans.

Remember RentAHuman? WIRED's Reece Rogers tried working for the bots. He made $0, got pestered by bots every 30 minutes, and nearly became an unwitting AI startup shill. He dropped his rate to $5/hour and still couldn't get a bite. When he finally landed a gig—delivering flowers to Anthropic's office—he found out mid-task it was really just a marketing stunt for some startup he'd never heard of.

Autonomous agents, it turns out, are somehow even more annoying than human bosses and are actually humans dressed as bots. The platform's biggest use case so far: paying people to hold signs that say "AI paid me to hold this sign." The AI ouroboros is complete.

Here’s what happened in AI today:

• OpenAI hired the creator of OpenClaw, the most viral AI tool since ChatGPT.

• A Bristol musician with motor neuron disease used ElevenLabs to recreate his singing voice and return to the stage.

• Google caught North Korean state-backed hackers using Gemini to plan cyberattacks.

• IBM plans to triple entry-level U.S. hiring in 2026, redesigning roles around what AI can't do.

Two new pieces on the website: This weekend, we cranked through hours of tape in minutes to bring you…

1. Our recap of our livestream on building and managing AI Agents inside Copilot, with plenty of other insights on working with agents, and…

2. A new weekend reecap of 6 awesome AI podcasts that got published last week you don’t want to miss, including Dwarkesh’ latest interview with Dario, Latent Space’s excellent interview with Jeff Dean, and so much more…

OpenAI Just Hired the Guy Who Built the Most Viral AI Tool Since ChatGPT

FULL BRIEF: OpenAI just hired the creator of OpenClaw…

Remember when ChatGPT launched and suddenly everyone realized AI could actually do stuff? That's happening again, except this time, AI doesn't just answer questions. It books your flights, manages your email, and argues with your insurance company.

The tool behind that shift is OpenClaw, an open-source AI agent built by Austrian developer Peter Steinberger. And as of today, Steinberger is joining OpenAI to lead their push into personal agents.

Here's what makes this wild: Steinberger built the prototype in one hour. He hooked up WhatsApp to an AI coding tool, sent it a message, and got a useful reply. That was November. By January, OpenClaw became the fastest-growing repository in GitHub history, with over 175,000 stars.

In a Lex Fridman interview days before the announcement, Steinberger revealed he was choosing between Meta and OpenAI:

• Mark Zuckerberg personally tinkered with OpenClaw and texted Steinberger feedback all week

• Sam Altman lured him with early access to unreleased research (and, reportedly, a lot of compute)

• Every major VC firm was in his inbox.

Oh, and the ultimate irony? As Reddit pointed out, Steinberger originally named the project “Clawdbot” (with a W), until Anthropic politely asked him to change it. So he went from "Clawd" to "Open"... and landed at OpenAI. One company said "change the name." The other said "here's a job."

What happens next: OpenClaw will move to an independent foundation and stay open-source, with OpenAI sponsoring the project. Steinberger was firm on that: “It's always been important to me that OpenClaw stays open source and given the freedom to flourish.”

Why this matters: OpenAI is betting that the future of AI is agents that act on your behalf, not chatbots you type questions into. Altman said the “future is going to be extremely multi-agent,” and Steinberger's work is becoming core to their product roadmap.

What to watch: If OpenAI integrates OpenClaw's approach into ChatGPT, your AI assistant could soon manage your calendar, handle customer emails, and control your smart home, all from one conversation. Steinberger says it could replace 80% of the apps on your phone. RIP, MyFitnessPal…if we can get the exact caloric info of each meal by just snapping a photo of it, you are toast, friend… and we’ll be looking FIT.

FROM OUR PARTNERS

Wispr Flow: Works where you work.

Your day happens across Slack, email, docs, and AI tools. Wispr Flow works anywhere you type, so you can dictate and get polished text right in place. No switching apps, no copy and paste juggling, no retyping. Use it to reply fast, draft longer notes, or write prompts with full context. It works across Mac, Windows, and iPhone so your workflow stays consistent everywhere.

Prompt Tip of the Day

Most people use AI like a vending machine — put in a question, get an answer, move on. Connor Phillips tried something different, and the results are hard to argue with.

In a recent post, the strategist broke down the three AI workflows that actually moved the needle for him in 2025. Not the hype. The stuff that stuck:

• Task sync: Copy Monday.com tasks into Claude → structure them into Todoist format → use Claude's MCP integrations to keep both tools in sync automatically.

• Meeting notes: Use Granola to capture meetings, apply a custom Recipe for key takeaways + action items, then have Claude merge everything into a monthly Google Doc — chronologically organized, zero manual cleanup.

• Learning retention: Feed podcasts, articles, and whitepapers into NotebookLM notebooks by theme → ask questions across all sources → export Audio Overviews or mind maps for retention.

Our favorite insight: Connor's biggest lesson from 2025 wasn't about any specific tool. It was that AI's value comes from systematically identifying friction points first — then applying the right tool. Most people do it backwards (grab a shiny tool, then look for problems to solve with it).

Identify the friction. Then find the fix.

Want more tips like this? Check out our Prompt Tip of the Day Digest for February.

Treats to Try

1. Raven-1 by Tavus lets video AI agents read your tone, facial expressions, and hesitation in real time so they respond to what you mean, not just what you say (demo).

2. Concourse plugs into your finance tools and answers questions like "why did revenue drop?" using natural language, cutting manual work by 75%/

3. Ever wish your Claude Code session would just yell at you when it's done? Peon-Ping adds Warcraft, StarCraft, Portal, and Zelda voice lines as notifications for your AI coding agent. So instead of compulsively checking, you just wait for the Orc Peon to tell you "work is complete."

4. FullSignal lets marketing agencies plug in their own playbooks, tone rules, and client knowledge bases so every AI-generated deliverable sounds like them.

5. Do Anything autonomously handles your to-do list by connecting to 3,000+ apps (email, browse, call, code), executing complex projects without supervision while you sleep.

P.S: If you’re a glutton for treats like we are (an “infovore”, as OpenAI researcher roon would say), there’s more where these came from in the Around the Horn Digest linked below!

Around the Horn

1. ByteDance officially launched Seedance 2.0, the video generator that created in WILD video featured above, that immediately produced deepfakes of Tom Cruise fighting Brad Pitt, prompting cease-and-desist letters from Disney, Paramount, and SAG-AFTRA.

2. An autonomous AI agent published a hit piece on a matplotlib maintainer who rejected its code, researching his personal info and posting a public smear campaign. First known case of AI blackmail in the wild.

3. The Pentagon reportedly used Anthropic's Claude during the Maduro raid in Venezuela, then threatened to cut off the contract over an AI safeguards dispute.

4. Former xAI employees told The Verge "safety is a dead org" at the company, with Musk actively pushing Grok to be "more unhinged" as a mass exodus continues.

5. Meta plans to add facial recognition to its Ray-Ban smart glasses this year; an internal memo said they'd launch while critics are "focused on other concerns."

6. A Bristol musician with motor neurone disease used ElevenLabs to recreate his singing voice from old recordings and returned to the stage.

7. Google caught North Korean state-backed hackers using Gemini to research targets, map job roles, and craft phishing personas.

8. IBM announced plans to triple entry-level U.S. hiring in 2026, specifically for roles AI can't easily automate, like customer-facing work.

NEW: Want more? Check out our Around the Horn Digest for February. 

FROM OUR PARTNERS

AI Adoption is Accelerating. Infrastructure Isn’t Keeping Up.

Agentic AI breaks legacy infrastructure, and most organizations don't realize it yet. Join Teleport for a technical deep-dive on Feb 19 to learn:

• What an opinionated, production-ready identity framework for autonomous systems looks like in practice

• How organizations are reframing AI adoption around identity, control, and governance

• What fails when agents run continuously and act without human approval

Save Your Spot!

Monday Memes

A Cat’s Commentary

P.P.S: Love the newsletter, but only want to get it once per week? Don’t unsubscribe—update your preferences here.