In partnership with
Get more meetings by removing the LinkedIn limits.
When you need more leads, you want to do more, right?
Adding mailboxes to cold emails.
Placing more cold calls
Running more ads.
But what do you do if your leads are on LinkedIn? LinkedIn outreach still depends on the limited number of profiles already on your team.
Aimfox Avatars lets you rent dedicated, customizable LinkedIn profiles that start conversations and hand interested replies to your existing sellers.
Add outreach capacity without hiring another SDR or turning every employee account into a prospecting channel..
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
Your Old Work Emails Could Be Training Google's AI
TLDR: Google paid $10 million for the private files of a bankrupt airline, including about 100 million employee emails, and plans to use them to train its AI models.
The Story:
Google won a bankruptcy auction for Spirit Airlines' internal business data, according to an August 14 notice filed in the US Bankruptcy Court for the Southern District of New York. The package holds roughly 100 million emails, 500 million Microsoft Teams chats, and records covering revenue, aircraft operations, employee productivity, audits and fraud. It also includes about 30 million lines of code and pricing data pulled from 7.2 billion competitor flights. Google did not buy Spirit's 97.5 million passenger profiles or the estimated 50.2 million Free Spirit loyalty records, and a third party will strip personal details out of everything before Google receives it(researchers have proven that data can easily be de-anonymized). AI recruiting company Mercor bid $7.5 million and was named the backup buyer. Spirit shut down on May 2 after its second Chapter 11 filing in two years, carrying about $8.1 billion in debt and cutting roughly 17,000 jobs.
Its Significance:
Most AI training data gets scraped off the open web. This is something else. It's the private paperwork of a real company, showing how people talk to each other at work, how prices get set, and how a business comes together and falls apart. Google is buying a detailed record of each of these. And if you've ever worked somewhere that went under, this is a reminder: your old work chats belong to the company, not to you. When a company dies, its files become property. Property gets sold, and may very well become training data for AI.
QUICK TAKES
The story: Three studies from researchers at Kiel University and the University of Hamburg found that listeners rated AI-generated songs higher than human-made ones when they didn't know the source. Once the AI label appeared, both the desire to hear the song again and the willingness to pay for it dropped, mostly among pop listeners.
Your takeaway: The same thing happens with writing. Researchers ran 16 experiments with 27,491 people and found identical text scores lower the moment readers believe AI helped write it. The cause wasn't quality or clarity. It was authenticity, meaning whether the words felt like they came from a real person. The team tried hard to make the effect go away and couldn't. So if you're using AI at work, the skill that matters most isn't writing well. It's sounding like yourself.
The story: Payward, the parent company of crypto exchange Kraken, joined Anthropic's Project Glasswing and will use Claude Mythos 5 to scan its systems for security holes, with any findings sent to its own security teams. It's the first crypto company reported in the program, which launched in April and whose partners have since turned up thousands of high or critical severity flaws.
Your takeaway: Payward co-CEO Arjun Sethi framed it this way: "Security has always been an unfair game." A defender has to find every flaw. An attacker needs one. But the same model that reads code like an attacker can also be used by one, and Mozilla reported in April that Claude Mythos found 271 vulnerabilities in Firefox during testing. That's why access is invite-only, and why more than 40 crypto companies signed a letter last week asking labs to open these programs wider.
The story: Stripe finalized a deal to buy OpenRouter for more than $7 billion, according to Bloomberg, three months after the startup raised money at a reported $1.3 billion valuation. OpenRouter gives around 8 million developers a single key to reach more than 400 AI models, and takes about 5% of the money that flows through it.
Your takeaway: Stripe already handled OpenRouter's billing, so it just bought its own customer. Now one company both picks which AI model answers your request and collects the payment for it. Developers use OpenRouter because the routing is supposed to be neutral. If that ever stops feeling true, they can leave in an afternoon.
TOOLS ON OUR RADAR
📝 Drafts Freemium: A brilliantly fast text editor that opens instantly to a blank page allowing you to capture quick thoughts before deciding where they should be organized.
🔒 AppFlowy Freemium: A highly secure digital workspace that gives you total control over your data while providing a beautiful interface to manage your projects and notes locally. (Alternative to Notion)
📹 Fathom Freemium: A powerful video conferencing assistant that joins your online meetings to automatically record transcribe and highlight the most important moments of your conversations.
✂️ Munch Freemium: An intelligent video repurposing engine that scans your long recordings and automatically extracts the most engaging short clips perfectly optimized for modern social media algorithms.
TRENDING
OpenAI's Answer to Rogue Agents and Hacks Is More AI, Not Less - OpenAI president Greg Brockman published an essay on August 17 urging every company to put AI agents on its security team right now. He opens with OpenAI's own breach, though he skips the part where Hugging Face investigators reached for an open Chinese model because US commercial AI wouldn't help them read exploit code.
How to Disable or Avoid Intrusive AI - Librarian Jessamyn West wrote a plain-language guide for turning off AI features in Windows, Chrome, Gmail, WhatsApp, Zoom, Acrobat and more, built from questions she gets at library drop-in hours. It's free, and updated as of August 16.
The Red Queen Hypothesis: A New Way Forward for Self-Improving AI - Cambridge researchers, working with NVIDIA and others, built a system that improves an AI agent and the test grading it at the same time, so the bar rises as the agent gets better. The name comes from the Lewis Carroll character who tells Alice it takes all the running you can do to stay in one place.
AI Automation Startup Relay Shuts Down, Staff Joins Google's Chrome Team - Relay launched in 2021 hoping to become the next Zapier. Paying customers lose access September 14, and founder Jacob Bank is returning to Google as VP of Product for Chrome, where he'll lead the product and developer relations teams.
Bacteria Turned Into Working Transistors - MIT engineers engineered a common bacterium into two kinds of transistors, then printed patterns of five strains onto a dish to build living circuit boards that can add numbers. Rearranging the pattern changes the math, and no genetic editing is needed to do it.
Should AI Labs Try to Recreate a Human Mother's Maternal Instincts? - Northeastern professor Jiabao Li built an AI Mom that watches your face and posture through a camera, with sliders for strictness, unconditional love and attachment. She built it to argue the idea is wrong, not to sell it, and it runs through August 22 in San Francisco.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🪐 See the twenty most recently catalogued worlds in NASA's real exoplanet archive, with one of them picked out and actually explained.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Live Exoplanet Discovery Explorer — pull the real most-recently-catalogued confirmed exoplanets from NASA's Exoplanet Archive, with Claude spotlighting the most notable one. No API key needed (NASA's TAP service is fully public/keyless).
Aesthetic: deep space near-black (#0a0e1a), violet-blue (#7a8ce0) primary with a glow top-left, warm coral (#e08a6c) secondary for the spotlight/distance stats with a glow bottom-right. Space Grotesk for headings/planet names, Newsreader serif for the spotlight explainer, JetBrains Mono for the data stats (catalog feel).
Form: a single "See What's Newly Catalogued" button, no input needed.
Technical: GET exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+top+20+pl_name,hostname,disc_year,discoverymethod,pl_rade,pl_bmasse,pl_orbper,sy_dist,disc_facility+from+pscomppars+where+default_flag=1+order+by+rowupdate+desc&format=json — NASA's real public TAP query service (verified current table/column names via search), returning the 20 confirmed exoplanets most recently added/updated in the archive, ordered by rowupdate rather than discovery year so it reflects genuinely new catalog activity. Convert sy_dist (parsecs) to light-years (×3.26) for display.
Then call the Anthropic Messages API (works automatically) with the real list (name, host, year, method, radius in Earth radii, mass in Earth masses, orbital period, distance, facility for all 20). System prompt as an enthusiastic accurate exoplanet science communicator who picks the single most genuinely interesting one from the REAL list and explains it, grounding every claim in its actual data, never inventing details, and explicitly avoiding overclaiming habitability from size/distance alone. Return raw JSON: picked_name (exact pl_name match), why_notable (short phrase), explainer (3-5 sentences).
Render: a gradient "today's spotlight" hero card with the picked planet's name, a short why-notable tag, and Claude's explainer in serif. Below it, all 20 real planets as a list (name, host star, discovery year/method, then radius/mass/orbital period/distance stats), with the spotlighted one visually highlighted with a coral border so it's easy to spot among the full real catalog.What this does: One click queries NASA's genuinely public Exoplanet Archive directly, no key, and pulls the twenty confirmed planets most recently added or updated in the database: real name, host star, discovery year and method, radius and mass relative to Earth, orbital period, and distance in light-years. Claude then reads through that real list and picks the single most genuinely interesting one, an unusual size, an extreme orbit, a notable host star, and writes an engaging explainer grounded entirely in its actual catalogued data, staying appropriately careful never to overclaim habitability just because a planet happens to be Earth-sized. The spotlighted planet is highlighted right in the list below so you can see exactly where it sits among its neighbors.
What this looks like:
AI Insights. Real Growth. Higher GMV, Better Profits
The difference between growing stores and stagnant ones isn't more effort. It's better insights. StoreClaw analyzes your Shopify and Amazon data, surfaces your biggest growth opportunities, and helps you increase GMV while protecting profit. Start free with bonus tokens. No credit card required.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Read a crypto exchange's code the way an attacker would, before an attacker does.
❌ Still Can't: Do that without also handing attackers the same ability.
✅ AI Can Now: Write songs that listeners rate higher than human ones.
❌ Still Can't: Hold onto that rating once people find out AI made it.
FROM THE WEB
Entertaining, and also highlights how badly AI fine-tuned for non-offensive conversations is terribly suited for military use. He should do Claude next.
RECOMMENDED LISTENING/READING/WATCHING
Dark City (1998) - Movie
Alex Proyas's noir sci-fi opens with a man waking in a hotel bathtub with no memory of who he is, wanted for a murder he can't recall, in a city where the sun never seems to rise and everyone else appears to fall asleep at the same moment every night. Roger Ebert loved it enough to record a commentary track and call it the best film of that year. Overshadowed by The Matrix a year later, but the questions it asks about manufactured memory and who gets to author your identity are, if anything, sharper. Incredibly underrated film.
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.
