- The AI Pulse
- Posts
- 🤖 Major Internet Publishers Make AI Pay To Crawl
🤖 Major Internet Publishers Make AI Pay To Crawl
PLUS: Fixing What Everyone Thought Was Unfixable
James Roesser
September 12, 2025
Welcome back AI enthusiasts!
In today’s Daily Report:
📚Major Internet Publishers Make AI Pay To Crawl
🚨Fixing What Everyone Thought Was Unfixable
🛠Trending Tools
🥪Brief Bites
💰Funding Frontlines
💼Who’s Hiring?
Read Time: 3 minutes
🗞RECENT NEWS
AI & LICENSING
📚Major Internet Publishers Make AI Pay To Crawl
Image Source: Canva’s AI Image Generators/Magic Media
Major internet publishers, including Reddit, Quora, and Yahoo, recently introduced “Really Simple Licensing (RSL),” which prevents AI Crawlers from gathering content from websites without consent or compensation to train advanced AI models.
Key Details:
RSL is embedded into websites to set up three distinct options for publishers to enforce on AI Crawlers:
🟢 Allow: Allow access to certain content.
🟡 Charge: Require to pay for certain content.
🔴 Block: Deny access to content entirely.
RSL relies on Fastly to develop a framework where only AI firms that agree to the rules, licenses, and regulations set by publishers are permitted to access their websites.
Think of it like a bouncer at a nightclub, where RSL checks the IDs and Fastly lets them in or turns them away.
Why It’s Important:
In June 2025, Anthropic’s crawl-to-referral ratio was 73,000:1. For every 73,000 web pages scraped by Anthropic’s AI Crawlers, only one person visited the original website where the information came from.
Publishers are attempting to create a scalable marketplace where both sides win. In theory, AI firms can achieve legal clarity through simplified licensing agreements, and websites can get a scalable way to monetize their content when it’s used to train advanced AI models.
🩺 PULSE CHECK
Has AI ever confidently presented false information as fact to you?Vote Below to View Live Results |
AI RESEARCH
🚨Fixing What Everyone Thought Was Unfixable
Image Source: Thinking Machines Lab CEO Mira Murati/“But what happens when we add two floating-point numbers with different exponents, such as 1230 and 23.4?”/Screenshot
Thinking Machines Lab just published “Defeating Nondeterminism in LLM Inference,” revealing the real reason why LLMs generate different responses to the same question.
Key Details:
Computers store information as binary data: a distinct sequence of 0s and 1s. This works great for whole numbers. For example, 100 in binary is 1100100.
But LLMs don’t deal with whole numbers; they work heavily with decimal numbers to calculate the likely next word within a sentence. For example, if you type: “The cat sat on the {BLANK}!” the LLM might predict “mat” → 0.81, “roof” → 0.13, and “moon” → 0.06.
To store these decimal numbers, LLMs rely on Floating-Point Arithmetic (FP), which is similar to scientific notation but in binary. But here’s the catch: a simple decimal number like 0.81 doesn’t have an exact FP. It’s like trying to convert 1/3 into decimal form; you get 0.3333.... repeating forever, so you must round somewhere!
As a result, FP isn’t perfectly precise, with tiny rounding errors baked into every calculation. We often blame these tiny rounding errors for causing LLMs to generate different outputs even when given the same input.
It turns out that LLMs are unpredictable primarily because the number of inputs they process at once changes constantly, not because of tiny rounding errors caused by FP.
Why It’s Important:
Imagine ordering the same cup of coffee every day at a coffee shop, but it tastes different depending on how busy the barista is. Why?! Because the barista has to juggle multiple coffee orders at once.
That’s exactly what’s happening with LLMs. For example, when ChatGPT handles a large number of user queries or prompts simultaneously, it processes them in batches for efficiency. However, this batching can change the quality of how your prompt is processed.
PROMPT ENGINEERING TIPS
⚙️Cultivating a Growth Mindset!
The world can be a competitive place. To keep up, it’s not just about working harder; it’s about working smarter.
With a growth mindset, you can steadily reach new heights in your professional journey simply through adjusting the way you think.
In other words, a growth mindset promotes the idea that abilities and intelligence can be developed with time, effort, and dedication.
Cultivating a growth mindset is crucial because it fosters resilience, facilitates learning, and fuels success.
This simple prompt turns ChatGPT into your personal growth mindset mentor:
Context: I want to build confidence in my ability to learn new things,
Clarity: but I often fall into fixed thinking, where I assume my abilities are limited.
Guidance: Can you help me develop a growth mindset regarding {Insert Specific Situation}, and explain how to redefine failure as a valuable stepping stone on my path to {Insert Specific Goal}?
I want to build confidence in my ability to learn new things, but I often fall into fixed thinking, where I assume my abilities are limited. Can you help me develop a growth mindset regarding {Insert Specific Situation}, and explain how to redefine failure as a valuable stepping stone on my path to {Insert Specific Goal}?🛠TRENDING TOOLS
👷Solid: if you can dream it, you can build it.
🔥Klap turns videos into viral shorts in minutes.
🎓Oboe creates magical courses to learn anything.
🎨PicturetoDrawing converts pictures to drawings in seconds.
🧠YouMind: an AI-based creation studio where learning meets writing.
🧰 Browse our Always Up-To-Date AI Toolkit.
🥪BRIEF BITES
ElevenLabs introduced “Voice Remixing,” which allows you to change the age, gender, or accent of any voice.
Stability AI unveiled “Stable Audio 2.5,” which enables brands to generate enterprise-grade sound effects for marketing campaigns.
The FTC issued “Section 6(b) Orders” to seven AI firms to seek information on how they test, measure, and monitor the potentially negative impacts of consumer-facing AI chatbots on children.
OpenAI officially secured Microsoft’s blessing to restructure OpenAI LP into a Delaware Public Benefit Corporation (PBC) that’s still controlled by a non-profit Board of Directors (BofD), marking the end of months of heated negotiations.
💰FUNDING FRONTLINES
You.com landed a $100M Series C at a $1.5B valuation to deploy AI-powered employees for enterprises.
Perplexity reportedly raised a $200M Series D at a $20B valuation for an AI-powered answer engine.
Replit closed a $250M Series C at a $3B valuation to turn ideas into apps.
💼WHO’S HIRING?
📒FINAL NOTE
FEEDBACK
How would you rate today’s email?It helps us improve the content for you! |
❤️TAIP Review of The Day
“Great information!”
REFER & EARN
🎉Your Friends Learn, You Earn!
You currently have 0 referrals, only 1 away from receiving 🎓3 Simple Steps to Turn ChatGPT Into an Instant Expert.
Share your unique referral link: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER