Welcome to Wednesday’s Newsletter

In today’s scoop 🍨 

• 🔄 Trump’s AI Deregulation Rollercoaster

• 🌐 OpenAI's Operator: The New Sheriff in Your PC?

• 🚀 DeepSeek’s R1: A New Challenger to OpenAI’s Crown?

• 🔧 3 Trending AI Tools

🔄 Trump’s AI Deregulation Rollercoaster

Yesterday, Donald Trump officially took office (again) and wasted no time revoking Joe Biden’s 2023 executive order on artificial intelligence. What does this mean? It’s deregulation season, baby. And AI is the headliner.

What Went Down

Trump has rolled back nearly 80 Biden-era executive orders, including one that required developers of advanced AI systems to share safety test results with Uncle Sam. This was part of a broader plan to tackle the risks AI poses to national security, public health, and safety.

Key highlights:

• 🔍 Biden’s order focused on safety testing: Developers had to provide transparency on systems that could endanger national security or critical sectors.

• 🔺 Trump’s priority: Unleash innovation. Guardrails? Not so much.

• 🛡️ Trump’s team isn’t offering immediate alternatives: There’s no new regulation to replace Biden’s policies—just vibes for now.

🚫 Safety First? Not So Much

Biden’s AI executive order had provisions to:

• Create the U.S. AI Safety Institute to establish voluntary guidelines.

• Address cybersecurity, chemical, and biological risks related to AI.

Trump’s scrapping of this order means safety standards are no longer mandated. It’s a wild west out there, with the potential for major breakthroughs — or major catastrophes.

Meanwhile, the EU is over here with its AI Act, banning facial recognition and regulating "high-risk" AI applications. America? It’s taking the “innovation above all” approach.

🔎 What It Means for AI’s Future

This hands-off strategy could:

• 🚀 Supercharge innovation: With fewer hoops to jump through, tech companies could push boundaries faster.

• 🚫 Open safety loopholes: Without transparency, risks like biased algorithms, deepfakes, and untested generative AI models could escalate.

“AI development rooted in free speech and human flourishing” is the Republican slogan, but some argue that human flourishing may take a hit if things spiral out of control.

🖼️ The Big Picture

Trump’s AI reset is like taking the training wheels off a rocket. It’ll either soar to unimaginable heights or crash spectacularly. Either way, it’s clear: The next few years will redefine the global AI race. Hold onto your algorithms.

🌐 OpenAI's Operator: The New Sheriff in Your PC?

Imagine a world where your PC does the work for you. Enter OpenAI's Operator—the AI tool that might just be the superhero (or supervillain, depending on your perspective) of productivity. Leaks suggest Operator is set to debut this month, promising to take the hassle out of multi-step tasks and maybe even book that dream vacation for you.

What We Know So Far

🔒 The Leak: Tibor Blaho, an AI product leaker with a knack for accuracy, uncovered evidence of Operator lurking in OpenAI’s macOS ChatGPT app. Think "Toggle Operator" and "Force Quit Operator" options hidden in plain sight.

🔄 Tasks It Could Tackle: From setting up spreadsheets to booking flights, Operator aims to handle tasks that typically make us groan. But before you daydream about outsourcing your entire to-do list, there are some hurdles…

🕛 Not Perfect Yet: Leaked benchmarks show Operator struggles with certain tasks:

• 🏋️‍♂️ Success Rate:

  • Signing up for a cloud provider and launching a virtual machine? 60% success.

  • Creating a Bitcoin wallet? A humbling 10%.

• 🔮 Benchmark Scores: Operator lags behind humans (72.4%) on OSWorld with a 38.1% score, but it’s leading the pack on web navigation tasks like WebVoyager.

⚙️ Challenges & Potential

OpenAI has reportedly focused heavily on safety, which might explain the tool’s long development cycle. Operator’s safety chops are decent, scoring well on tests designed to provoke it into dodgy behavior. However, rivals like Anthropic have already thrown their hats in the ring, and the AI agent race is heating up.

🎡 Market Projection: The AI agent market could hit $47.1 billion by 2030. It’s a gold rush, and Operator might just be OpenAI’s ticket to striking it rich.

🔎 The Big Question

Will Operator be the assistant of your dreams or another overhyped AI tool? As we wait for OpenAI’s official announcement, one thing is clear: If it works, Operator could redefine how we interact with our devices.

🚀 DeepSeek’s R1: A New Challenger to OpenAI’s Crown?

If OpenAI’s o1 model is the reigning champion, DeepSeek’s new reasoning model, R1, might just be the scrappy underdog ready to shake things up. The Chinese AI lab recently unleashed R1 into the wild—and it’s already making waves.

🔬 Benchmark Smackdown

• 🔢 AIME 2024 (math puzzles): R1 scored a whopping 79.8%, narrowly outshining o1’s 79.2%.

• 🌍 MATH-500: R1 hit 97.3%, beating o1’s 96.4%.

• 💻 SWE-bench Verified (coding tasks): R1 scored higher than its rivals, showcasing serious programming chops.

R1’s secret weapon? It’s a “reasoning model” that fact-checks itself and solves problems with meticulous thought. Sure, it takes a little longer, but the reliability is unmatched, particularly in domains like math and physics.

🥟 More for Less Here’s the jaw-dropper: R1 costs a staggering 90-95% less than OpenAI’s o1. For businesses and developers, this is a game-changer:

• 💳 $0.55 per million input tokens versus OpenAI’s $15.

• 💸 $2.19 per million output tokens versus OpenAI’s $60.

Plus, the open-source model is available on Hugging Face under an MIT license, meaning it’s free for commercial use and tinkering.

🌐 A Global AI Power Struggle DeepSeek’s move comes amid escalating U.S.-China tensions in AI. OpenAI’s VP of Policy has singled out DeepSeek’s parent company as a “concern,” while stricter export controls could complicate the development of Chinese AI systems. Still, R1’s release signals that China’s AI labs are stepping up—fast.

🔑 Takeaway

DeepSeek’s R1 isn’t just another AI model; it’s a statement. With stellar benchmarks, lower costs, and open-source accessibility, R1 might be a wake-up call for industry giants.

🔧3 Trending AI Tools

• 🚀 Milestone Content Studio - All-in-one AI platform that boosts marketing velocity by automating workflows, brainstorming strategies, and executing campaigns seamlessly.

• 🔥 /extract by Firecrawl - AI-powered tool for extracting structured web data with a simple prompt, delivering clean JSON for lead enrichment, KYB automation, and no-code workflows in seconds.

• 📝 Spellar AI - Native macOS and iOS meeting assistant that supports 100+ languages, capturing conversations and exporting seamlessly to Notion, Google Docs, Linear, Jira, and Confluence.

Thank you for reading!

Did something in today’s newsletter catch your eye, make you laugh, or spark a new idea? Hit reply and let us know—we’re always up for some feedback (or a friendly chat).

And if you’ve got a friend who’d love this AI goodness, don’t keep it to yourself. Share the scoop and make their inbox a little smarter. 🚀

Until tomorrow—stay curious! 👋