Sponsored by
Moda is the AI design agent with taste
Moda's viral launch hit 4.4 million views in two days. Tens of thousands of professionals signed up. Startups, agencies, forward-thinking brands and top firms are now using Moda to create brand-aligned slides, ad creative, reports, social carousels and more.
Most AI tools tend to create what we call "AI slop": repetitions of the same colors, layouts and fonts. And when you try to fix it, you get stuck in a loop of re-prompting.
Moda is different. Drop in your website URL, and Moda learns your brand from the ground up: your colors, your fonts, your visual language. Then it helps you generate pro-quality slides, docs, and marketing assets.
The best part? Every layer is fully editable on a real canvas, and exports to powerpoint, PDF and more.
Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
OpenAI Shut Down China-Linked Bots Pushing Fake Posts About Data Centers and Tariffs
TLDR: OpenAI says it caught accounts linked to China using ChatGPT to write fake posts meant to turn Americans against AI data centers and tariffs, but the effort mostly flopped.
The Story:
OpenAI put out a new threat report this week. In it, the company said it shut down a set of accounts, likely based in China, that used ChatGPT to stir up bad feelings about data centers among regular Americans (OpenAI says it caught China using ChatGPT). The accounts fell into two groups. One pushed posts and images claiming data centers were driving up electricity bills for families. The other attacked U.S. tariffs. The operators pretended to be Americans, but they typed their prompts in simplified Chinese, and they even had ChatGPT draw comic strips to post with fake accounts. OpenAI thinks the group was a social media team at a Chinese company doing work for the government. China can't officially use ChatGPT, so the accounts snuck in with VPNs. The campaign didn't catch on, and OpenAI found no sign it spread far past the fake accounts themselves.
Its Significance:
The fake posts didn't invent new lies. They picked fights people were already having. Electricity really does cost more near some data centers. So the goal was to pour gas on a fire that was already burning. One strange thing stood out: China has its own AI models, but this group used an American one instead, and even OpenAI said it couldn't explain why. Expect a lot more interactions with AI-enhanced bots in the forums and comment sections of your favorite publications and social media platforms.
QUICK TAKES
The story: For years, Pokémon Go asked players to scan real-world spots with their phones. A Dutch report found that about 30 billion of those scans were used to train a system that helps machines figure out where they are without GPS. Niantic's spatial company later teamed up with a defense contractor to put that tech into military drones and robots that work where GPS is jammed, like war zones.
Your takeaway: One player told reporters, "First you think you are playing a game, and then suddenly your data can be used in a war." That's the risk with free apps. The data you hand over for fun can get sold and reused for things you never agreed to, years later. When you scan, snap, or share inside an app, it helps to assume that data could end up far outside the app one day.
The story: Coinbase launched "Coinbase for Agents," a tool that lets AI assistants like ChatGPT and Claude connect to your account and trade crypto, make payments, and manage what you own, all within limits you set. It's the second big money move this week. Mastercard just rolled out its own system that lets AI agents pay each other, with more than 30 partners signed on, including Coinbase, Stripe, and OKX.
Your takeaway: If you already know how to trade and you set tight limits, handing some tasks to an AI agent could save you real time. If you're new and don't fully understand the trades, this can go very wrong very fast. An AI moving real money while you're not watching can drain an account before you catch the mistake. Big companies are racing to give AI access to your money right now, so the limits you set matter more than ever.
The story: Anthropic released Claude Fable 5, its most capable public model yet. Buried in a 319-page safety document was a rule that quietly made the model give weaker answers on certain advanced AI research, without telling the user. After researchers pushed back, Anthropic apologized and changed it, so the model now says out loud when it's handing your request to a less powerful backup.
Your takeaway: This is the tightrope every top AI company is walking. They want to show they have the most capable models. They also don't want those models helping people do dangerous things, like building risky AI or weapons. Anthropic's fix keeps the safety limit but makes it visible, so you know when you're not getting the full model. Whether that's the right balance is something the whole field is still arguing about.
TOOLS ON OUR RADAR
📰 FreshRSS Free and Open Source: A lightweight and customizable news aggregator that collects articles from all of your favorite blogs and news sites into one single feed so you can stay informed without being tracked by social media algorithms.
📂 FileBrowser Free and Open Source: A user friendly web interface that turns any folder on your computer or server into a private cloud storage system where you can upload download and manage your files from any device with a web browser.
💬 LibreChat Free and Open Source: A unified chat interface that brings all of your favorite artificial intelligence models together into one highly customizable and professional platform.
🔐 CryptPad Free and Open Source: A privacy focused collaborative office suite that allows you to work on documents spreadsheets and presentations with others in real time using end to end encryption to ensure only you can see your work.
TRENDING
Jeff Bezos' AI Startup Prometheus Hits $41 Billion - Bezos's new company raised $12 billion to build what he calls an "artificial general engineer," AI meant to speed up designing and building physical things like jet engines and medical devices. He says a change that takes engineers 10 years today could be cut by a factor of 10 or more.
ElevenLabs Launches Avatars for Talking-Head Videos - ElevenLabs added a tool that turns a written script into a lip-synced talking-head video in one step, pairing its AI voices with video. It's aimed at marketers and teachers who want to make lots of videos fast without juggling separate apps.
OpenAI Weighs a Price War With Anthropic as China Undercuts Both - OpenAI is thinking about big price cuts for developers to compete with Anthropic, but Chinese models like DeepSeek already cost a fraction of either. One report found the same job costing $4,811 on Claude versus $544 on a Chinese model.
Deezer's Free Tool Scans Spotify and Apple Music for AI Songs - Deezer built a free tool that checks your playlists on Spotify, Apple Music, and around 18 other services for fully AI-made tracks. The company says it now gets about 75,000 AI songs a day, roughly 44% of all new uploads.
Judge Cancels Trial After Lawyers on Both Sides Used AI - A Mississippi federal judge threw out a trial and fined four lawyers after both sides filed briefs with fake, AI-made case citations. Two of them were banned from her court for two years.
Anthropic Apologizes Over Claude Fable 5's Secret Censorship - Anthropic said sorry after people found its new Fable 5 model was quietly weakening some answers without telling users. The timing was awkward, since the company had just asked rivals to slow down on frontier AI while its own product slowed down the rivals using it.
Apple Stock Slides After Its AI Reveal at WWDC 2026 - Apple showed off a smarter Siri and more on-device AI at WWDC, but the stock fell about 8% over the week because investors wanted a firm launch date and bigger news. Siri AI won't launch in the EU or China at first, two markets worth about 35% of iPhone sales.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
⚒️ Paste any facts, names, or sequences. Get custom acronyms, rhymes, and image-hooks that make them stick.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Mnemonic Forge — turn facts/names/sequences into acronyms, rhymes, and image-hooks. Persist to localStorage key 'mnemonic_forge_v1'.
Aesthetic: dark forge brown (#16100a), molten orange glow gradients at bottom. Fraunces serif for headings and italic, Space Grotesk for body, Space Mono for labels. Fire orange (#ff8c3c) primary with gradient button, plus per-technique accent colors: orange=acronym, cyan=rhyme, purple=image, green=story, gold=association, pink=chunking. Hammer-swing animated loading icon. Dashed-border device showcase boxes.
Form: content textarea (raw paste), type dropdown (ordered sequence / unordered facts / names / vocab / numbers / process / framework), style dropdown (variety / mostly acronyms / images / rhymes / stories), optional personal-hooks input.
System instructions to the model: act as master mnemonist using real techniques memory champions use (acronyms, method of loci, vivid bizarre images, rhymes, peg systems, chunking, personal association). Brain remembers absurd, vivid, emotional, personal. Make devices genuinely sticky, weave in personal hooks. Return raw JSON: target_summary (1 sentence), mnemonics array (3-5 items: type acronym|rhyme|image|story|association|chunking + for + device (THE actual mnemonic, vivid, with **bold** anchors and *italic* emphasis) + how_to_use + why_it_works (start with **bolded principle name**)), best_bet (2-3 sentences picking which to lead with, **bold** the technique), drill array (4 recall drills with timing using spaced recall).
Render: orange "what we're making stick" target card. Mnemonic cards each with type-colored left border, type badge + "for X" label, large showcased device box with dashed border (bold/italic rendered), how-to-use section, why-it-works footer with bolded principle. Gradient "forge master's pick" card. White drill card with numbered 48-hour steps. Markdown bold/italic parsed throughout.What this does: Drop in whatever won't stay in your head — an ordered list, a set of facts, people's names, vocab, numbers, a process. Pick the type, the mnemonic style, and any personal hooks (interests, humor, references you love). Get back 3–5 forged memory devices using the real techniques memory champions use: acronyms, vivid bizarre images, rhymes, story-journeys, and chunking. Each shows the actual device, how to deploy it on recall, and the cognitive reason it works. Plus the forge master's pick of which to lead with and a 48-hour drill to lock it in. Saves to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Connect to your crypto or bank account and make trades and payments on its own, inside limits you set.
❌ Still Can't: Tell its own real sources from fake ones. It keeps citing court cases that were never written, and lawyers keep getting caught.
✅ AI Can Now: Turn a written script into a lip-synced talking-head video in a single step.
❌ Still Can't: Make people actually believe a message at scale. China's bot posts were polished, but OpenAI found they barely spread past the fake accounts.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
The Electric State by Simon Stålenhag - Book
A 2018 illustrated novel set in an alternate 1997 America where a war between humans and AI drones has just ended and the country is littered with the rusting corpses of giant machines. A teenage girl drives across this broken landscape with a small yellow robot in the passenger seat, looking for her brother. Stålenhag's paintings are some of the most haunting AI imagery ever put on paper, and the whole book reads in about an hour.
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
What if AI found your next job overnight?
Stop wasting hours filling out repetitive applications. AIApply automatically discovers relevant jobs, optimizes your resume, generates personalized applications, and applies for you around the clock so you can focus on securing your next opportunity faster.