In partnership with
Welcome back to The AI Field. Ask Claude to remake a game you love, and it may now say no. Anthropic quietly added a rule against copying specific existing games, and confirmed it on Friday.
Also inside: Claude's top plan loses its five-hour limit, and a Google AI chip wrote a poem in orbit. Plus six tools and a five-minute way to cut your phone and internet bill.
In today's AI Field:
Claude now refuses to remake other people's video games
Claude's $200 plan is now Ultra, with no 5-hour limit
Google ran its AI chips in space for the first time
Plus: six tools, four quick hits, and a five-minute way to cut your phone bill.
Read time: 5 minutes
Nobody's coming to approve your idea
People assume that real businesses get built by other people with more resources. It's how great ideas get stuck in notes apps and conversations.
But you don't need six months and a developer's salary to validate your concept. You just need a weekend. Describe what you want to exist and Lovable builds it: a real site, a real product, ready to accept real payments from real customers. Not a mockup to show people. An actual thing they can use.
You don't need anyone's sign-off to get started. Take $5 off your first month.
Discount of $5 applies to first month of any paid plan. You need to sign up via this webpage in order to claim the discount. The discount applies to new and returning users without an active paid plan. Terms apply.
LATEST DEVELOPMENTS
ANTHROPIC · CLAUDE CODE
Image: chasm on X.
The AI Field: Claude Code, Anthropic's coding tool, picked up a new two-sentence rule this week: don't rebuild a specific existing game, even if it's renamed or called original. Addy Osmani, who works on Claude Code, confirmed it on Friday.
The details:
The rule: "Do build original games, games in a genre, and games from the user's own designs and assets. Don't build a recreation of a specific existing game, even if it is renamed or called original." A user spotted it around October 27. It comes from Anthropic's servers, and there's no setting to turn it off.
What still works: original games, games in a genre, like a new platformer or puzzle game, and games made from your own designs and assets. Osmani says it was added for Opus 5.5 in Claude Code, and asks people to send him cases where Claude turns down something that's theirs.
Why: using Claude to infringe someone else's intellectual property was already against Anthropic's usage policy. The rule just makes Claude act on it.
Why it matters: AI can now rebuild a game in an afternoon, which is hard on the people who made the original. But Claude can't check who owns what, and one small studio says its paid work for clients who own the games got blocked too. Anthropic says it will keep adjusting the line. If you use Claude for your own game, nothing should change.
TRY THIS TODAY · 5 MINUTES
Cut your phone and internet bill with one message
One person had his AI agent read his phone and internet bills and write one message to the provider, which he sent himself. The company waived a switching fee, removed a forgotten Disney+ add-on and cut both mobile plans, plus a 15% loyalty discount. He puts it at about EUR 469 a year, which we haven't verified. He used Grok Bot, but any free AI chat that reads files and searches the web can do this.
1. Download your last three phone and internet bills as PDFs or screenshots.
2. Attach them to your AI chat with this prompt:
"Read my last three phone and internet bills [ATTACH]. List every line, plan, add-on and fee with what I pay each month, and flag anything I may not use. Then check the provider's current prices for similar plans. Draft one short, polite message to the provider asking them to remove the extras, move me to a cheaper plan and tell me about any loyalty discount. Don't send it. Show me the draft and your estimate of the yearly savings first."
3. Check the draft and send it yourself. Ask about lock-in before saying yes to a new deal, and check your next bill.
ANTHROPIC · CLAUDE PLANS
Image: Anthropic.
The AI Field: Anthropic also renamed its two top Claude plans on Friday and removed the five-hour usage cap on the most expensive one. Prices stay the same.
The details:
New names: Max 5x is now just Max at $100 a month, and Max 20x is now Ultra at $200 a month.
The catch: the five-hour limit is gone only for verified Ultra accounts, starting with most existing subscribers. Newer accounts keep a temporary limit until Anthropic verifies them automatically from account and payment history. Adding a phone number doesn't count.
Still capped: Ultra keeps a weekly limit of 10 times Pro, a separate weekly limit for Fable, and other caps may apply. To check your account, open Settings, then Usage. If there's no "Current session" bar, you're verified.
Why it matters: The five-hour wall was one of the most hated limits in paid AI. Anthropic, which just filed to go public, is betting heavy users will pay $200 a month to work without being cut off. If you were on Max 20x, you don't need to do anything.
Stop typing AI prompts. Start talking.
You think 4x faster than you type. So why are you typing prompts? Wispr Flow turns your voice into ready-to-paste text inside any AI tool. Speak naturally, tangents and all, and Flow cleans it up. Available on Mac, Windows, iPhone, and Android.
GOOGLE · AI IN SPACE
Image: Google.
The AI Field: Google says the test satellite for Project Suncatcher, launched on October 1, gave the first AI answer from a data-center-class chip in orbit. Its free Gemma 4 model wrote a haiku, which was sent down to a ground station near Svalbard.
The details:
The poem: "Blue marble below / Data streams in silent grace / Starlight feeds my core."
The hardware: the satellite carries four of Google's TPU AI chips, but only one runs at a time, for about 20 minutes, on up to 1 kilowatt of power.
What's next: tests of how the chips handle heat and radiation, then two connected satellites in 2027 to see if chips in orbit can work together.
Why it matters: AI data centers need huge amounts of power and land, and towns near them are pushing back. In orbit, the sun shines almost all the time. One haiku is a long way from a data center, but it shows Google's AI chips can run up there. Google, which launched Gemini 4 Carbon this week, is planning where its future models will run.
TRENDING AI TOOLS
📝 Gemini: You can now create and edit files right in the chat. Ask for a doc, spreadsheet, slides or code, preview it and ask for changes. PDF editing works on web and mobile, the rest on the web. Free for everyone.
🍝 Zest: Text Zest like a friend who knows where to eat. It picks spots from where you and your friends actually go and can book the table, even from a group chat.
💬 Knocket AI: Add live chat to your website with one line of code. Upload your FAQs, and its AI agent answers routine questions and hands harder ones to you, even on Telegram.
⚙️ Make: Make's Anthropic Claude app runs Claude inside your workflows, for example to handle each new row in a Google Sheet and write the answer back. Needs a Claude API key. Free plan, no card needed.
🧱 Base44: Want to sell lessons, bookings or a small product? Describe the site or app you want, and Base44 builds it with payments included, so customers pay you directly through it. Free plan to start.
✨ Genspark: Need a deck by Monday? AI Slides now runs Gen-1, Genspark's own model trained just for slides, to turn one prompt into a full first draft. On every plan, with free daily credits.
QUICK HITS
📉 AI stocks had a rough Friday: Intel fell 25% after a weak forecast, Micron dropped 5.5% and Nvidia 4%. A popular AI stock fund is down 8% in October.
🛵 Almost nobody orders DoorDash through AI agents: DoorDash said outside AI agents have placed only "several hundred" orders in total, in a quarter with about a billion orders.
🧒 ChatGPT gets stricter on violence and weapons: OpenAI is tightening how ChatGPT talks about past mass attacks for everyone, and weapons info for teens, over the next few weeks. Its first Community Safety Report is due in early 2027.
🚫 AI labs agree on one opt-out for AI training: OpenAI, Google, Microsoft and Anthropic will honor one robots.txt line, __usage_ai-training, that tells AI companies not to train on a site. Anthropic has started, and the others follow by 2027.
That's it for today!
See you soon,
Olle Hellman | Founder of The AI Field