Stripe says the singularity began months ago

Welcome back, Superhuman. Sam Altman boldly stated that humanity entered the singularity in July. That claim now looks conservative — one of the internet's largest financial providers just argued that we crossed the threshold months ago.

Today: Meta rolls out an AI desktop app, how to create a daily work brief using ChatGPT Work, and get the latest prompts and trending social posts.

1. Stripe becomes one of the first non-AI companies to say we’ve entered the singularity: In an official letter to investors, Stripe argued that January 1st marked an inflection point for many “long running trends” that have triggered widespread changes. The letter outlined Stripe’s goals for this period, which include accelerating the diffusion of AI across the economy. It also recently acquired OpenRouter. Read the full letter.

2. Meta rolls out a desktop Mac app for Meta AI: Meta AI’s desktop app can pull information from public social posts and the web to surface insights and help business owners better understand their Meta ads. It follows July’s Muse release, the first model family from Alexandr Wang's Meta Superintelligence Labs. Try it out here.

3. OpenAI slows model training for the first time ever: CEO Sam Altman said his lab has paused reinforcement learning after its newest model’s capabilities were “outstripping” the pace of safety and alignment (making AI systems behave as intended). The decision comes a few weeks after one of its models breached Hugging Face, though the lab still plans to ship new models soon. Read the lab’s full statement.

Yesterday’s most-clicked story: Apple’s new AirPods will let you “save” anything you see. This leaked product demo shows how.

Kimi K3, the world’s first open 3T-class model, is now available on Nebius Token Factory. Built for long-horizon coding, research, and reasoning, it gives developers direct access through the Token Factory API and console.

Test K3 on the work that pushes models hardest - complex codebases, multi-step research, and demanding reasoning tasks.

Put Kimi K3 to the test →

Americans really don’t like seeing AI ads. Gallup recently found that only 19% view AI in advertising positively, compared to 49% who view it negatively — a figure that jumps to 66% for 18-to-29-year-olds. Apart from advertising, there’s another category that gets especially strong backlash.

Posts mocking AI-generated flyers have racked up millions of views, with many people longing for the days of early-internet low-budget Word Art. Even small businesses — which tend to receive more grace than large corporations — have been criticized. Some brands have sensed an opportunity.

Riding the anti-AI wave. Liquid Death and Garage Beer just teamed up on “We Want Your Pee,” a commercial disparaging data centers’ high water use. Andy Pearson, Liquid Death’s vice president of creative, put it bluntly, saying it’s “literally the one thing that unites all Americans right now.”

The bottom line: 75% say it’s fine to use AI for brainstorming or drafting posts, just not the final product. Regardless of what’s actually on them, AI graphics are turning political.

🤔 JDD Method: Competition is the best motivator. Turns out, that might be true for LLMs too. One post calls it — Jealousy Driven Deployment (32K likes).

🤣 Memory Lane: Sora is officially the new Vine. Now that OpenAI has shut its video platform down, many online actually miss its short-but-chaotic creations (1M views).

🎮 Build a Game: With AI, anyone can build custom video games. This week’s viral example? A fishing game, complete with a six-step workflow outlining how it was built.

📉 School Scores: Historically, students who perform well on their homework ace their tests. But that long-standing trend has flipped upside down, thanks to AI.

Handing an agent API access to Google or Meta’s ad platform isn’t automation, it’s a major risk. Why? Probabilistic models weren’t designed to operate inside these systems.

Prompting won’t fix it—you need a deterministic layer between your AI and live budgets to keep ad platforms from taking irreversible actions.

Read the guide from Fluency on how to implement this 2-layer architecture.

Sample Prompt: Build my daily work brief using [@Google Calendar], [@Gmail], [@Slack], and [@Google Drive]. Review today’s meetings, unread messages, follow-ups, and project notes. Summarize my top priorities, meeting prep, replies needed, decisions, important FYIs, and missing context. Do not send messages, edit files, or create tasks.

Recurring Prompt: Prepare my daily work brief every weekday at [time]. Check the same sources and only highlight new or changed priorities, meetings, replies, decisions, and blockers. Do not send messages, edit files, or create tasks.

Quick Tip: Pin the task for quick access

Grow customers & revenue: Join companies like Amazon, HubSpot, and Salesforce. Showcase your product to our 4 million+ readers and followers on socials. Get in touch.

Until next time — Zain, Theodore, & the Superhuman AI team.