Anthropic is giving away billions in equity
Plus, how to use the ChatGPT council prompt, DeepSeek releases a low-cost AI model and web search tools for agents, and more!
Hola Decoder😎
If someone forwarded this to you and you want to Decode the power of AI and be limitless, then subscribe now and Join Decode alongside 30k+ code-breakers untangling AI.
💰 Anthropic pledges 5% of its equity to fund public benefits
Anthropic has introduced the Anthropic Public Benefit Fund, a proposed initiative that would allocate 5% of the company's equity, worth approximately $48 billion at its May valuation, toward addressing AI's societal impact.
The Shift:
Anthropic Sets Aside $48 Billion - Subject to shareholder approval, the fund would receive 5% of Anthropic's equity through a company-owned LLC. It could support nonprofits and direct financial distributions to specific populations.
Funding Could Grow With Anthropic's Value - For every 50% increase in company valuation beyond an undisclosed baseline, Anthropic's board could allocate another 2% of equity to the fund, linking potential public benefits to company growth.
AI Job Disruption Becomes a Priority - Anthropic identified AI's impact on workers as a key funding area, building on its earlier $200 million commitment to research employment disruption and evaluate possible policy responses.
20% of Shares Earmarked for Public Good - Anthropic says approximately 20% of its shares are designated for public-benefit purposes, combining the proposed fund, its co-founders' 80% giving pledge and company-matched employee donations.
Anthropic is proposing a way to connect the financial gains from AI development with support for people affected by the technology. The fund could provide substantial long-term resources, although its initial allocation requires shareholder approval and future increases remain discretionary.
Together with The Shift
Stop Reading About AI. Start Getting Results From It.
AI is everywhere, but practical, actionable ways to use it are really hard to find.
The Shift fixes this with a focus on a single topic per edition. A real problem AI can solve today.
You’ll see real examples, short workflows, and proven prompts that help you work faster, think better, and automate intelligently under 5 minutes a day.
Once you subscribe, you get access to 3,000+ curated AI tools and 1000+ prompts, so nothing stays theoretical.
If you want AI to work for you in real, practical ways, not just ideas, you’ll love The Shift.
🤹How to use the ChatGPT council prompt
ChatGPT agrees with you by default. The Council Prompt puts five advisors in the room who don’t.
Step 1: Paste the Council Prompt - Open a new chat and paste this:
“For any decision or idea I bring you, assemble a council of five advisors: a Devil’s Advocate who finds every flaw, a Realist who checks the facts, a Strategist who thinks three moves ahead, an Expert in the relevant field, and a Chairman who synthesizes all four views into a final verdict. Each advisor speaks in turn before the Chairman rules. Do not skip any advisor or merge their voices.”
Step 2: Bring your real decision - Drop in whatever you are actually trying to decide. Business idea, job offer, major purchase, life move. The more specific the better.
“I am considering [decision]. Run it through the Council.”
Step 3: Push back on the verdict - Tell the Council what you disagree with and why. Watch which advisors update their position based on new evidence and which ones hold firm. The ones who hold firm are usually pointing at the real risk.
Step 4: Save it in custom instructions - Go to Settings > Personalization > Custom Instructions and paste the Council Prompt. Every important conversation now runs through five perspectives instead of one agreeable voice.
Use it for any decision you are excited about, excitement is exactly when you need the Devil’s Advocate most.
🚀 DeepSeek releases a low-cost AI model and web search tools for agents
DeepSeek has released V4.1-Flash-1017, an open-weight AI model designed for affordable coding tasks, alongside new Search and Fetch tools that let AI agents access live web information at significantly lower prices.
The Shift:
DeepSeek Challenges Leading Coding Models - The updated model scored 77.0 on DeepSWE v1.1, ahead of Claude Opus 5 (74.0) and GPT-5.6 Sol (73.0) in DeepSeek's own tests. However, Opus 5 still leads on broader reasoning benchmarks.
AI Model Costs Drop Sharply - DeepSeek charges $0.15 per million input tokens and $0.60 per million output tokens off-peak. The model's weights are also available under an MIT license, allowing developers to run and commercialize it independently.
Web Search Tools Cost 85% Less - DeepSeek's new Search and Fetch tools cost $0.75 per 1,000 successful requests, compared with $5 or more for some competitors. Both tools are temporarily free through supported coding agents until November 1.
China Proposes New AI Export Restrictions - Beijing has proposed controls covering 23 AI-related items, including certain models exceeding 2 trillion parameters or ranking among the world's top 10. The draft could affect distribution of advanced Chinese AI models.
DeepSeek is increasing pricing pressure on AI providers by combining competitive coding performance, openly available model weights and inexpensive web tools. While premium models retain advantages in difficult reasoning tasks, lower-cost alternatives could become increasingly attractive for routine coding and high-volume AI agents
Together with Playbook Pro
Your Growth Strategy Is Costing You. Here’s the Math
Ad costs are climbing. Most first-time buyers disappear within months, and margins are shrinking fast when discounts drive sales.
We spent 200+ hours on research, distilling what top brands do into a field guide you can read and implement in under 30 minutes.
Inside Rewriting the DTC Rulebook, you’ll learn how to:
📈 Lift retention by 5% to unlock 25–95% profit growth
🔥 Launch emotional campaigns proven to double repeat sales in 8 weeks
💰 Use the Brand Value Canvas to raise prices 15–40% without losing volume🧠 Slash decision cycles from 3 weeks to 3 hours using the Assumption Buster Toolkit
📊 Cut paid-ad dependence by 30% in 60 days with the Owned Media Checklist
🏆 Tools you Cannot Miss:
🎬 ngram 2.0 – Turns prompts, documents, websites, and recordings into explainer videos, product demos, and narrated content with editable scripts, storyboards, captions, and voiceovers.
✂️ OpenVids – Edits existing footage using text instructions, automatically assembling rough cuts, adding motion graphics, adjusting audio, and reviewing the final video on an editable timeline.
📸 Ephora – Removes unwanted objects, text, and backgrounds from photos, generates professional headshots, and creates branded product images directly on mobile.
🛠️ PocketWebTools – Processes images and audio directly on your device, including background removal, image upscaling, subject cutouts, transcription, and text extraction without uploading files.
🎥 Refs – Helps video creators provide structured visual references for AI-assisted video workflows, making it easier to communicate the intended style and direction
🚀 Quick Hits
🏛️ Democrats are considering tougher oversight of AI companies ahead of the U.S. midterm elections, focusing on data centers, job displacement, and safety risks. Gaining congressional committee control could enable hearings, investigations, and demands for company records.
🤖 A Philadelphia woman was arrested and charged with disorderly conduct after refusing to move from an Uber delivery robot’s path. She said she was protesting automation replacing human delivery workers, highlighting growing concerns over job displacement.
🚫 Major websites are blocking AI agents from booking services, shopping and accessing accounts. Yelp, Delta, and major banks cite licensing restrictions and security concerns.
📰 Meta removed terms suggesting it could scrape news websites for AI training following backlash. The company says its crawling practices remain unchanged and websites can still use robots.txt.
🤖 Instagram head Adam Mosseri says the platform is exploring how to handle AI-generated fakes copying creators’ content. He says banning AI is impractical, while non-copyrighted material presents challenges.
🤳AI Nugget of the Day
Thanks for Decoding with us🥳
Your feedback is the key to our code! Help us elevate your Decode experience by hitting reply and sharing your input on our content and style.
Keep deciphering the AI enigma, and we'll be back with more coded mysteries unraveled just for you!