Welcome back, Superhuman. Eleven days ago, Moonshot AI debuted Kimi K3: a new frontier-level model so popular the lab had to temporarily pause new subscriptions. Today, they’ve made it available for anyone to download and run themselves. For free. Frontier intelligence is now available to anyone — not just the big labs and those who pay for their products.
Today: Sam Altman heads to Washington, how to turn Reddit posts into viral content with AI, and get the latest prompts and trending social posts.
TODAY IN AI
Nvidia is partnering with Ilya Sutskever’s Safe Superintelligence. Image source: Reuters.
1. Moonshot AI releases the full weights for Kimi K3: The model weights and technical paper for Kimi K3 are now available for research, personal use, and commercial deployment (with guardrails), making it the largest open weight model ever released. K3 debuted eleven days ago and experienced such high demand that the lab suspended new subscriptions. If this adoption continues, it will increase pricing and performance pressure on top US labs. Learn more about Kimi K3.
2. Nvidia backs a stealth AI lab, launches an open-source alliance: Nvidia is investing up to $5B in Safe Superintelligence (the secretive startup founded by former OpenAI chief scientist Ilya Sutskever) after the lab confirmed it has research that’s “worth scaling.” Nvidia is also heading the newly-formed Open Secure AI Alliance, a coalition focused on promoting open-source AI as a cybersecurity asset. The initiative comes amid growing concern over AI’s cyber risks. Read more from the WSJ.
3. Sam Altman to meet with senior government officials: OpenAI’s CEO is heading to Washington this week to preview the lab’s most powerful model — the one that played a key role in hacking Hugging Face last week. The Trump administration is still finalizing its official regulatory stance on AI, but signed an executive order in June establishing a voluntary 30-day pre-release window to vet new models. Read more.
Yesterday’s most-clicked story: Anthropic published an Opus 5 Prompting Guide that outlines strategies and best practices for using its new model.
SPONSORED BY SPINACH AI
Spinach AI is the most accurate AI notetaker in 100+ languages with official Claude and ChatGPT connectors.
It records, transcribes & summarizes meetings, then:
- Feeds Claude, ChatGPT or Cursor with the context they need
- Syncs action items to Linear, Jira, Monday, Asana, Trello or ClickUp
- Updates your CRM - HubSpot, Salesforce, Zoho or Attio
- Posts summaries to Slack, Teams, Notion, Confluence, Google Docs
Backed by Y Combinator, Zoom & Atlassian. Trusted by 500,000+ professionals.
Set up Spinach in 30 seconds →
FROM THE FRONTIER
The internet’s click economy is dying and major publishers are starting to push back
Made with Midjourney
The rise of AI Overviews. Google’s AI Overviews are flipping the internet’s content ecosystem upside down. Last year, these overviews appeared in just 15% of searches. But now? They show up in 43% of searches, answering queries before people even visit a website. The flip has led to decreased traffic for major publishers, and now many are rethinking their relationships with the world’s largest search engine.
Publishers push back. Publications including Reuters, USA Today, Politico, and People are debating cutting off Google’s access to their content, according to a WSJ report. The thinking: if Google is no longer sending web traffic their way, why let it scrape their content for free? Mike Reed, CEO of USA Today, said, “It’s time to take a stand and say enough is enough.”
Reddit is also having second thoughts. Reddit and Google signed a $60M annual licensing deal in 2024 that let the search giant train its AI models on Reddit threads. But Reddit reportedly may not renew. Executives are “assessing what the upside is” of continuing to feed content to Google, implying Reddit hasn’t seen much value from the deal.
Where do we go from here? AI helps us get answers to our questions faster than ever before. But most of its answers are sourced from an information supply chain — the internet at large — that relies on ad-supported clicks. If that base layer starts to erode, it’ll eventually take the AI layer with it.
IN THE KNOW
What’s trending on socials & headlines today
Meme of the day
🎤 Unlocking Productivity: An AI power user boldly claimed that he works twice as efficiently on-the-go using ChatGPT Voice than at his computer — the post has sparked a debate about work/life balance in the AI era (4M views)
👀 Landing the Role: If you’re on the job hunt, try using Claude as a real-time interview partner. One Redditor says doing mock interviews (out loud) made the real interview feel like a rerun (2.5K upvotes).
🎮 Claude of Duty: Leading frontier models continue to push the boundaries of what regular people can build with AI. The latest example: A professional-grade shooter game (4M views).
🤦 Facepalm: A Senator is going viral for accidentally reading an AI prompt as part of his speech. One commenter says they got a physical injury from cringing so hard (3.5K upvotes).
💭 Agent Insight: A viral post suggests that you should avoid using agents to “rediscover the same solution” over and over again. Agents should only ever handle exploration. (12K likes).
SPONSORED BY GLEAN
Gartner’s new Emerging Market Quadrant shows how enterprise AI is moving beyond pilots into governed, production-ready systems, and Glean is recognized as a Market Shaper. For leaders evaluating AI platforms, the takeaway is clear: useful agents require more than ease of use.
PRODUCTIVITY
🤖 Kit For AI: Convert PDFs and web pages into clean Markdown for LLM workflows.
📗 ContentIQ: Verify facts before writing articles with cited sources.
💼 StaffingLeads: Get real-time hiring leads for recruiters.
TUTORIAL
How to turn Reddit posts into viral content with AI
Find a relevant Reddit post in your niche.
Add /.json to the end of the URL and copy the thread data.
Paste it into ChatGPT to extract pain points and generate viral hooks.
ChatGPT Prompt: "[add 3 viral posts] [add JSON file] Analyze this Reddit JSON. Extract the biggest pain points and write 5 viral two-line hooks that match the style of my top-performing hooks."
Send the hooks and your best-performing post samples to Claude, if you have any.
Claude Prompt: "[add 3 viral posts] Using these viral [platform] posts as references, write 5 new posts on this topic with the same structure, pacing, formatting, and voice, without copying the wording."
Claude will write 5 posts that match your voice and proven format.
PROMPT
Product Advertising
ChatGPT Image 2.0: Ultra-realistic luxury product advertising campaign, a futuristic white architectural room with tall geometric walls and a glossy black-and-white checkered reflective floor. A giant premium over-ear headphone in the foreground (black matte body with deep burgundy ear cushions), positioned as the hero product using forced perspective. Behind it, a stylish young male fashion model wearing black sunglasses, a cream bomber jacket, beige shorts, and white sneakers crouches dynamically, reaching one hand toward the camera with exaggerated perspective. Clean luxury lighting, soft volumetric glow, floating dust particles, subtle audio waveform graphics in the background, minimal premium aesthetic, editorial fashion photography, commercial product render, ultra-sharp details, cinematic composition, depth of field, 8K, Octane Render, photorealistic. Leave space at the top for branding and the headline "Hear Your Style" in elegant modern typography.
EXTRAS
Want more?
Check out our Top 125 AI Tools
Get better at prompting with our Top 1,000 Prompts
Learn how to use AI at work with our 50 Tutorials
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.
What did you think of today's email?
Until next time — Zain, Theodore, & the Superhuman AI team.
