📜OpenAI 15-page love letter

In partnership with

🧠 Your Daily Dive Into the Wild World of AI
📅 March 14, 2025

Good morning, AI enthusiasts! ☕ Whether you’re sipping coffee or letting your robot butler brew it 🤖, we’ve got a packed lineup of today’s hottest AI stories. From OpenAI’s regulatory power play to Google’s personalization glow-up, it’s all here—served up with a side of wit and emojis. Let’s dive in! 🚀

The future of presentations, powered by AI

Gamma’s AI creates beautiful presentations, websites, and more. No design or coding skills required. Try it free today.

🛡️ OpenAI’s Big Ask: Federal Shield or Bust

OpenAI wants Uncle Sam to take the wheel on AI regs

📜 OpenAI just dropped a 15-page love letter to the White House as part of its AI Action Plan, and the TL;DR? They’re begging for a federal shield to fend off a chaotic swarm of state-level AI rules—781 bills this year alone! ⚖️ In return, they’re offering to cozy up under Washington’s oversight.

🔍 Here’s the scoop: OpenAI says this patchwork of state regs could kneecap U.S. innovation 🇺🇸, especially with China’s AI machine humming along, fueled by “unfettered access to data.” 🇨🇳 They’re pushing for:
💰 Infrastructure cash
📖 Copyright reform (think fair use for AI training)
📊 Wider access to government datasets
⛔ A ban on “state-controlled” models like DeepSeek, calling them security risks 🚨

🔥 Why it matters: With their $500B Stargate Project 🌌 and a growing D.C. fanbase, OpenAI’s not just building tech—they’re shaping policy. But asking for state-law immunity while throwing shade at open-source rivals? That’s a tightrope walk 🎭, especially with their own closed-source drama and copyright lawsuits piling up. 👀

🌟 Gemini Gets Personal: Google’s AI Levels Up

Your search history just became your AI’s BFF

🔮 Google’s Gemini AI assistant is getting a glow-up 💅 with new personalization features, tapping into your Search history (and soon, apps like Photos and YouTube 🎥) to serve up responses so tailored they might know you better than your mom. 😳

🛠️ The deets: Powered by the snazzy Gemini 2.0 Flash Thinking model ⚡, this opt-in feature lets the AI peek at your digital footprint 👣 to boost relevance. Think:
🧗‍♂️ “Hey, you searched for hiking trails last week—want some gear recs?” 🎒
🎭 Free users also get Gems (custom chatbots) and Deep Research tools, previously locked behind the Advanced paywall. 🔓

🔐 Privacy-first folks can disconnect anytime, and it’s 18+ only. Smart move, Google! 👏

🔥 Why it matters: Google’s sitting on a goldmine of user data 📊, and they’re finally flexing it to make Gemini a standout. It’s a high-stakes game—nail the balance of personalization and privacy, and they’ve got a winner 🏆. Mess it up, and it’s a trust-busting PR nightmare. 😬 Either way, we’re watching! 👀

⚡ Cohere’s Command A: Lean, Mean, Enterprise Machine

Big AI power, tiny GPU footprint

🚀 Cohere just unleashed Command A, an enterprise-grade AI model that’s fast, efficient, and ready to take on the big dogs—all while sipping power from just two GPUs. 🎛️ Yep, you read that right.

⚙️ The rundown:
⚡ 156 tokens per second—1.75x faster than GPT-4o and 2.4x speedier than DeepSeek-V3
🧠 Matches or beats them in human eval for business, STEM, coding, and agentic tasks
📜 256k context window and fluency in 23 languages 🌍
🏢 Enterprise perks: advanced RAG + integration with Cohere’s North platform for secure, in-house AI agents

🔥 Why it matters: While others chase benchmark bragging rights, Cohere’s betting on efficiency—delivering top-tier AI that doesn’t need a data center to run. 🏭💸 For companies obsessed with cost and security, this could be the golden ticket 🎫. Watch out, big spenders! 💰

🔍 Other AI Headlines You Should Know

• 🌐 Google secretly owns 14% of Anthropic, with its investments in the OpenAI rival now exceeding $3B.

• 📚 AI2’s OLMo 2 32B just became the first open model to outperform GPT-3.5 and GPT-4o mini while using a fraction of the compute.

• 🎥 Moonvalley’s Marey AI video model claims to be trained only on licensed content, making it a potential Hollywood-friendly option.

• 📼 Perplexity’s new Model Context Protocol now lets Claude tap into real-time web searches for fresher responses.

• 🎮 Xbox’s new AI assistant, Copilot for Gaming, is coming to mobile soon, offering in-game coaching and social enhancements.

• 📱 Alibaba’s New Quark AI assistant app just got a major upgrade, now running on its new Qwen reasoning model.

• 🎬 Google is integrating YouTube into Gemini AI, allowing the model to analyze and interact with video content.

• 👩‍💻 Former OpenAI researcher Andrej Karpathy predicts AI will reshape web design, optimizing pages for LLMs over human readability.

💻 Mastery tools of the day

• Harvey - AI for law firms, service providers, and Fortune 500 companies

• R1-Omni - Alibaba’s new open-source multimodal reasoning model that can ‘read’ emotions

• Reka Flash 3 - Open, 21B parameter reasoning AI for on-device deployment

💡What else are we reading and seeing?

• HTTP/3 is everywhere but nowhere

• China's AI frenzy: DeepSeek is already everywhere — cars, phones, even hospitals

• Nvidia's next chips are named after Vera Rubin, astronomer who discovered dark matter

That’s it for today’s Not So Artificial! Thanks for reading—see you in the AI future. 🚀

Learn how to make AI work for you

AI won’t take your job, but a person using AI might. That’s why 1,000,000+ professionals read The Rundown AI – the free newsletter that keeps you updated on the latest AI news and teaches you how to use it in just 5 minutes a day.

Sign up to start learning.