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
Mythos AI That Was "Too Dangerous" to Release Is Coming Out Soon
TLDR: Anthropic says its most powerful AI model, Claude Mythos, will reach all customers "in the coming weeks," just months after the company warned it was too dangerous for a public release.
The Story:
Anthropic said Thursday it expects to bring Claude Mythos to all its customers soon, once it finishes building more safety guards. Mythos is built for cybersecurity work, and it's good at it. The U.K.'s AI Security Institute said the model finished a complex 32-step fake attack on a company network all on its own. Mozilla said it found 271 separate bugs in the Firefox browser. Up to now, only a small group of partners like Microsoft, Apple, and Nvidia could touch it through a locked-down program called Project Glasswing. Anthropic first called Mythos "the most powerful AI model we've ever developed" back in March, and warned it could find and break into software faster than human defenders could keep up.
Its Significance:
Anthropic spent months telling everyone Mythos was too risky to hand out to everyone, then is handing it out anyway, not too long after. OpenAI's Sam Altman called this "fear-based marketing," saying Anthropic talked up the danger to make the model seem more special. He might have a point. A tool that was truly too dangerous for the world doesn't usually become safe enough for everyone in a couple of months. The truth read is probably in the middle: Mythos is a real step up in cyber skills, and that matters for the people who patch software and the hackers who attack it. But the giant warnings also did Anthropic a favor by making the model sound legendary before anyone outside a few labs had used it. Claude is still the Beginners in AI preferred model, and is incredibly powerful on its own, even without all the extra hype.
QUICK TAKES
The story: Researchers at Shanghai Jiao Tong University and Tencent built an AI agent called ProAct that uses the quiet time between your messages to guess your next question and prep the answer early. In tests, it cut follow-up requests by 11.7% and made 28.1% fewer made-up answers, though no real people tried it yet.
Your takeaway: Most AI sits and waits for you to type. This one does homework while you're away. If it works outside the lab, your tools could start finishing your thoughts, which is handy and a little unsettling at the same time. Being that AI is a prediction machine, this is no surprise that we're seeing more of this.
The story: Anthropic raised $65 billion in new funding, pushing its value to $965 billion and passing OpenAI's $852 billion for the first time. The company says it's now pulling in $47 billion a year, up from $10 billion last year, mostly from businesses using Claude to write code.
Your takeaway: The maker of Claude is now the most valuable AI company in Silicon Valley, and it got there by selling to businesses, not chasing casual users. When a five-year-old company is worth almost a trillion dollars, the tool you use at work is probably here to stay.
The story: On the Pat McAfee Show, NBA boss Adam Silver said the league will move to an AI camera system for "objective" calls like out-of-bounds, similar to the Hawk-Eye tech used in tennis. Human refs would still handle judgment calls like fouls, since cameras can't always tell real contact from a flop.
Your takeaway: This is the line a lot of jobs will draw with AI. Let the machine handle the clear, measurable stuff, keep humans on the calls that need judgment. Fans are split, and the same fight is coming to plenty of workplaces.
TOOLS ON OUR RADAR
💰 MoneyManagerEx Free and Open Source: A visually intuitive and highly comprehensive personal finance application that helps you carefully track your bank accounts monitor daily expenses and organize your family budget offline.
📖 Sigil Free and Open Source: An incredibly useful publishing tool specifically designed to help aspiring authors easily format visually organize and seamlessly package their own electronic books for digital distribution.
📹 Jitsi Free and Open Source: A brilliant and completely free video conferencing platform that allows you to instantly host large virtual meetings with your friends and family without creating any user accounts.
📖 Sigil Free and Open Source: An incredibly useful publishing tool specifically designed to help aspiring authors easily format visually organize and seamlessly package their own electronic books for digital distribution.
TRENDING
The Internet Is Being Rebuilt for Machines - AWS and Cloudflare are redesigning core web plumbing because AI agents don't browse like people do. Agents spin up many helpers that hit hundreds of databases in seconds, so AWS launched a new search system built for that kind of traffic instead of steady human clicks.
Florida Supreme Court Amends Rules to Address AI Use in Court Filings - Florida now requires anyone signing a court document, lawyer or not, to certify that the legal cases they cite actually exist and are quoted correctly. The rule takes effect June 15 and comes after AI tools invented fake court cases in real filings. Penalties go up to having a case thrown out.
AI Warfare Is Here, and CBS News Got a Look at the U.S. Military Training to Use It - At the African Lion 2026 exercise, CBS saw a gun-mounted robot roll across the Moroccan desert and drones carrying explosives, all tested alongside AI software like Palantir's Project Maven. One officer said a targeting decision that once took hours took three minutes, with a human still approving the strike.
Perplexity Computer Comes to Word, Excel, PowerPoint, and Outlook - Perplexity's AI assistant now works inside four Microsoft Office apps through a side panel, drafting docs, building slides, updating spreadsheets, and writing emails. The company says Computer has done $4.4 billion in "labor-equivalent work" since launching under three months ago.
Introducing Dynamic Workflows in Claude Code - Anthropic added a feature that lets Claude Code plan a big job, then run hundreds of smaller AI helpers at once to finish it and check its own work. One engineer used it to move a 750,000-line codebase in 11 days while keeping 99.8% of tests passing.
Anthropic's Claude Opus 4.8 Is Here: Better AI Coding, Smarter Safety, Same Huge Price - Anthropic shipped Opus 4.8 just six weeks after 4.7, with better coding and a model that's about four times less likely to hide its own mistakes. Pricing stays at $5 and $25 per million tokens, which looks steep next to China's DeepSeek V4 Pro at under a dollar.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
📕 Describe one of your bad habits. Get the bestselling self-help book that gets written about overcoming it — fake blurbs included.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Self-Help Book Generator — full bestseller treatment for any bad habit. Persist to localStorage key 'selfhelp_book_v1'.
Aesthetic: deep navy background (#1a1a2e), gold (#f0c060) accent. Playfair Display heavy serif for book title and headings, Inter for UI/labels, Caveat for handwritten touches. Book cover with randomized gradient color theme (crimson/mustard/emerald/cobalt/onyx) and subtle shimmer animation. Stats strip in black with three big gold numbers. White section for TOC and excerpt. Trademark method shown as massive acronym letters.
Form: habit input, how-it-shows-up textarea, author archetype dropdown (reformed addict / ex-Wall Street meditator / Harvard academic / ex-Navy SEAL / Silicon Valley founder / Brooklyn life coach), genre dropdown (productivity / spirituality lite / behavioral science / memoir-meets-method / tough love / cosmic woo).
System instructions to the model: write the airport bookstore bestseller version. Slightly satirical but earnest. Title is punchy, often paradoxical. Method is trademarked acronym. Blurbs sound like famous-adjacent people. No em dashes. Return raw JSON: title, subtitle (longer "How to..." explainer), author_name + author_credentials, weeks_on_list / copies_sold / languages (stats), toc array (7 chapter titles + page numbers), excerpt_chapter + excerpt_title + excerpt_text (2-3 paragraph opener with one made-up stat or study), method_acronym + method_steps array (one per letter with name + desc), blurbs array (3 items: quote + attr_name + attr_credentials), tagline.
Render: full book cover with gradient background, BESTSELLER badge, big title, italic subtitle, gold divider, author + credentials. Black stats strip with three gold numbers. White TOC section with numbered dotted-line entries with page numbers. White excerpt card with red left-border. Method card with massive acronym letters + lettered step cards. Dark blurbs section with gold-star ratings. Centered gold italic tagline at bottom.What this does: Name a habit, describe how it shows up, pick an author archetype and genre flavor. Get the full book treatment: a randomized-color cover with title, subtitle, author, and credentials, bestseller stats (weeks on list, copies sold, languages), a 7-chapter table of contents, an actual Chapter One excerpt that reads like a real airport bookstore opener, a trademarked acronym method, and three glowing blurbs from famous-adjacent people. Plus a marketing tagline. Saves to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Find thousands of real software bugs on its own, like the 271 Firefox flaws Mythos turned up.
❌ Still Can't: Tell a real foul from a flop in a basketball game, which is why human refs keep the judgment calls.
✅ AI Can Now: Run hundreds of smaller AI helpers at once to finish a huge coding job and check its own work.
❌ Still Can't: Reliably promise its legal citations are real, which is why Florida courts now make a human swear to it.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
A novel told through five intertwined voices across four centuries, from a 17th-century Puritan girl's sea diary to letters by Alan Turing to a 2040 inventor in prison for making AI companion dolls that scored 90% on the Turing Test. Hall traces the entire genealogy of how we tried to teach machines to talk, and what happens to the kids who grow up loving them more than the humans around them.
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.
The LA Mayor Odds Are Moving. June 2 Is the Deadline.
Bass at 68%, Pratt at 27% — and $21M already trading on the outcome. The best trades happen before the picture gets clear. Get in now and get $10 free to start.
Trade responsibly.