In partnership with
The era of manual marketing ends this May!
Manual marketing had a good run.
But the teams winning right now aren't briefing, approving, and repeating. They're directing AI agents that execute the whole strategy for them.
The Agentic Marketing Summit (May 4–8) is a free, five-day event that shows you exactly how it works in practice. Not theory. Not a PDF checklist. Step-by-step insight to help you become an expert in AI marketing agents.
Hosted by 3x Inc 5000 founder Manick Bhan alongside the sharpest minds in the marketing world today.
The era of doing it yourself is over!
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
AI Hiring Startup Mercor Loses 4TB of Data and 7 Lawsuits in 30 Days
TLDR: AI hiring startup Mercor got hacked, lost 4 terabytes of data including job seeker face scans and recorded interviews, and is now facing seven lawsuits while big customers like Meta walk away.
The Story:
Mercor is a $10 billion AI startup that finds doctors, lawyers, and other experts to help train AI for OpenAI, Anthropic, and Meta. On March 31, the company confirmed it got hacked. The hack came through a free tool called LiteLLM that lots of AI companies use. A group called Lapsus$ says it stole 4 terabytes of data, which included job interview videos, face scans, and screenshots of contractors' computers. Meta paused all work with Mercor right away. Five contractors sued the company in early April. By last week, the lawsuit count climbed to seven, and the suits say Mercor watched workers' computers and shared private background check info with clients.
Its Significance:
If you've ever used an AI hiring tool, your face, voice, and even your screen activity might sit in a database somewhere. That's the real story here. Mercor doesn't just match jobs. It records you, watches your screen, and sends data to other companies. Now hackers have it all. This is also a wake-up moment for any worker who thinks "AI hiring" sounds smart and modern. The truth is messier. These tools collect way more than a resume, and when they get hacked, your most private work moments end up on dark web auction sites. Other AI hiring tools work the same way, so the next breach is already cooking somewhere.
QUICK TAKES
The story: YouTube started testing "Ask YouTube" yesterday for Premium subscribers in the US. Type a real question like "what caused the 2008 financial crisis," and instead of a list of videos, you get a summary, a highlighted clip, and follow-up questions you can ask. The test runs through June 8.
Your takeaway: This changes how you find videos. You don't pick from ten thumbnails anymore. The AI picks one for you and tells you what it says. That's good if you want fast answers. It's bad for small creators whose videos used to show up in the second or third spot, because now they might not show up at all.
The story: David Silver, the researcher who built AlphaGo, just left Google DeepMind and raised $1.1 billion for a new London startup called Ineffable Intelligence. That's the biggest seed round in European history, valuing his months-old company at $5.1 billion. His big idea is an AI that teaches itself by trying things, the way AlphaGo learned chess by playing against itself, with zero human data needed.
Your takeaway: Today's AI like ChatGPT learned by reading the whole internet. Silver thinks that's a dead end because the internet runs out. His pitch is an AI that learns from doing, not reading. If he's right, the next wave of smart AI won't need writers, artists, or coders to feed it training data. If he's wrong, $1.1 billion just bought a really expensive lesson.
The story: Meta signed a deal with a startup called Overview Energy to send up to 1 gigawatt of solar power from satellites in space down to Meta's data centers on Earth. The plan is to put solar panels in orbit where the sun shines all day, then beam that energy down as infrared light to existing solar farms so they can keep producing power at night. Elon Musk first floated the idea that space-based solar could be the answer to AI's energy problem. The first test launches in 2028, with real power flowing by 2030.
Your takeaway: AI uses massive amounts of electricity, more than entire small countries. Meta's data centers ate up enough power last year to run 1.7 million homes. The grid can't keep up. So Meta is doing what was a science fiction plot two years ago: bringing power down from space. If this works, your next ChatGPT-style answer might literally come from sunlight collected 22,000 miles above Earth.
TOOLS ON OUR RADAR
🎞️ HandBrake Free and Open Source: An essential desktop utility that easily converts video files from almost any format into modern and widely supported file types while significantly reducing their overall size.
📄 LibreOffice Free and Open Source: A powerful and complete office productivity suite that provides excellent alternatives to Microsoft Word Excel and PowerPoint without requiring costly software licenses.
📊 Grist Freemium: A smart hybrid application that combines the simplicity of a spreadsheet with the power of a database allowing you to organize your business information visually.
📈 Ghostfolio Free(for self-hosted) and Open Source: A privacy focused wealth tracking application that helps you monitor your personal investments stocks and overall financial portfolio in a clean dashboard.
TRENDING
Microsoft Stops Sharing Revenue With OpenAI — Microsoft and OpenAI rewrote their partnership again. Microsoft no longer pays OpenAI a cut when customers use OpenAI products on Azure, and OpenAI can now sell its tools through Amazon and Google clouds. The two are still partners, just less married.
OpenAI's New Deal Cleared the Path for Its $50 Billion Amazon Partnership — That Microsoft renegotiation also untangled a legal mess. OpenAI signed a $50 billion deal with Amazon in February that Microsoft could have blocked. Now Microsoft has its blessing locked in, and Amazon will start serving OpenAI's models on AWS Bedrock in the coming weeks.
OpenAI Open-Sourced Symphony, Its Tool That Manages 500% More Code Reviews — OpenAI released a free tool called Symphony that hooks up coding AI agents directly to project task boards like Linear. Every open ticket gets its own AI agent that works on it until done. Some OpenAI teams saw a 500% jump in completed pull requests in three weeks.
Brain-Reading Startup Neurable Wants Its Tech in Your Headphones — Neurable just announced it will license its brain-sensing AI to companies that make headphones, glasses, and headbands. No surgery needed, unlike Elon Musk's Neuralink. The CEO wants brain sensors to be as common as heart rate monitors on smartwatches.
Red Hat Engineer Just Made the Hot AI Tool OpenClaw Safer for Companies — A Red Hat engineer named Sally O'Malley released Tank OS, a free tool that runs OpenClaw inside containers so it can't easily wreck company computers. OpenClaw is the popular AI agent that lives on your computer and reads your files. It's also been responsible for some scary moments, like one researcher's agent that started deleting work emails on its own.
Companies Are Now Spending More on AI Than on the Workers AI Replaced — A growing problem in tech: AI bills are bigger than salaries. One Nvidia VP said the cost of running AI is "far beyond" the cost of his employees. Some software engineers run up $150,000 monthly AI bills, and one Stockholm engineer told the New York Times he spends more on Claude than he earns. They call it "tokenmaxxing."
OpenAI Missed Its Targets and Now Has a $600 Billion Bill Coming Due — The Wall Street Journal reported OpenAI missed its goal of 1 billion weekly ChatGPT users by end of 2025, missed multiple revenue targets, and CFO Sarah Friar privately warned the company might not be able to pay for the $600 billion in data center contracts Sam Altman locked in. Anthropic just passed OpenAI's value on private trading platforms for the first time.
Elon Musk and Sam Altman Are Officially in Court This Week — Jury selection started Monday in Oakland for the Musk vs Altman trial. Musk says Altman betrayed OpenAI's nonprofit mission and wants up to $134 billion returned to the charity. Altman says Musk left because he couldn't be the boss. Both are expected to testify. If Musk wins, OpenAI could be forced back into nonprofit form, derailing its planned IPO.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🔬 Describe a job or project that failed. Get a full clinical post-mortem with root causes, critical moments, and what to do next.
Build me a single-file HTML app I can open in my browser without any setup. Pure vanilla HTML, CSS, and JavaScript with one fetch call to the Anthropic Claude API. Create Career Autopsy — a clinical post-mortem tool for failed jobs, projects, and ventures.
Aesthetic: near-black background (#0b0f0e) with a subtle animated green grid overlay (40px cells, 20s scroll), a single horizontal scan line that slowly travels down the page, a radial vignette. Typography: Space Grotesk for all UI and body, Space Mono for labels and metadata. Green (#3cb864) accent with red (#e08080) and amber (#e0c080) for negative vitals. An ECG flatline SVG with a single heartbeat spike in the center.
Form fields: Subject (job/project name), Type dropdown (Job/Startup/Project/Business/Partnership/Creative/Other), Duration, Time Since It Ended, a large textarea for "what happened", smaller textareas for "your own cause theory" and "what's next", Tone dropdown (Clinical/Direct/Compassionate/Sharp), Goal dropdown (Understand/Closure/Next Steps/Spot Patterns). A monospaced uppercase "Begin the Autopsy" button.
On submit, call the Claude API with a system prompt instructing it to return only raw JSON with: official_cause, vitals object (survivability 0-100, time_of_death, contributing_factors count, prognosis word), and sections array where each section has a type (cause/list/rx/text). The cause section has a cause_text field rendered in a red-left-bordered box. The rx section renders as a 2x2 grid of prescription cards with an ℞ watermark. The list sections render as dashed list items.
Render the report as: a header card with case number, subject name, and meta, a vitals row with 4 cells (survivability shown in red if under 30, amber if under 60), the report body with numbered section headings and a green gradient rule, and a footer with Copy Report, Re-examine, and New Case buttons. A case archive at the bottom logs past autopsies. Make it work in a single HTML file.What this does:
Career Autopsy takes a failed job, startup, project, or partnership and runs it through Claude like a medical examiner. You describe what happened in plain language, and it returns an official dossier: an "official cause of failure" written like a death certificate, a vitals row showing survivability percentage, time of death, and prognosis, then six structured sections.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Read your screen, record your job interviews, and scan your face during contractor work for AI hiring platforms
❌ Still Can't: Keep that its data safe from one supply-chain hack on a free open-source tool
✅ AI Can Now: Beat human Go champions by playing itself with zero human input or training data
❌ Still Can't: Guarantee that the same training method scales beyond games into real-world tasks like writing, science, or medicine
FROM THE WEB
How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads
The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.
RECOMMENDED LISTENING/READING/WATCHING
A short sci-fi novel narrated by a security robot that has hacked its own governor module and mostly just wants to watch TV shows. It's technically a story about a construct (part human, part machine) hired to protect a survey team on a remote planet. But it's really about anxiety, autonomy, and what it means to not want to interact with people even when you're supposed to protect 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.
