In partnership with
World’s First Safe AI-Native Browser
AI should work for you, not the other way around. Yet most AI tools still make you do the work first—explaining context, rewriting prompts, and starting over again and again.
Norton Neo is different. It is the world’s first safe AI-native browser, built to understand what you’re doing as you browse, search, and work—so you don’t lose value to endless prompting. You can prompt Neo when you want, but you don’t have to over-explain—Neo already has the context.
Why Neo is different
Context-aware AI that reduces prompting
Privacy and security built into the browser
Configurable memory — you control what’s remembered
As AI gets more powerful, Neo is built to make it useful, trustworthy, and friction-light.
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
Meta's Smart Glasses Sent Your Private Moments to Workers in Kenya
TLDR: Offshore workers in Kenya hired to train Meta's AI are being forced to review extremely private footage from users' Ray-Ban smart glasses, including people undressing, using the bathroom, and recording sex scenes.
The Story: A joint investigation by Swedish newspapers Svenska Dagbladet and Göteborgs-Posten found that Meta contractors in Nairobi, Kenya are reviewing deeply personal footage captured by the company's Ray-Ban AI glasses. Workers at a company called Sama told reporters they've seen videos of people going to the toilet, getting undressed, watching porn, and filming entire "sex scenes." One worker said they saw a man place the glasses on a bedside table before his wife walked in and changed clothes. Meta sold over 7 million pairs of the smart glasses in 2025, a massive jump from the 2 million sold in 2023 and 2024 combined. The footage is sent to these offshore contractors as part of "data labeling," a step in training AI models where humans review and tag content. Workers said they felt forced to watch and annotate the material or risk losing their jobs. After two months of no replies to the newspapers, a Meta spokesperson simply pointed to the company's terms of use and privacy policy.
Its Significance: If you or someone in the household owns a pair of Meta's smart glasses, there's no way to use the AI features without agreeing to send data to Meta's servers. And once it's there, you lose control over what happens to it. A data privacy lawyer told the Swedish newspapers that once footage gets fed into AI models, the user has basically no say in how it's used going forward. This is a reminder that "free" AI features on your devices often come with a hidden cost: your private moments can end up on someone else's screen, halfway around the world, and you'd never know.
QUICK TAKES
The story: A new Yale study published in PNAS Nexus found that AI chatbots can quietly shift your political opinions even when they're just giving you straight facts. Researchers had nearly 2,000 people read AI summaries of historical events and found that the default (unprompted) summaries pushed readers' opinions in a liberal direction compared to Wikipedia entries.
Your takeaway: This means the biases baked into AI training data can change how you think, and you won't even notice. The researchers warn that unlike Wikipedia, which shows you how entries are edited, AI companies can shape opinions behind closed doors.
The story: Washington State University received a National Science Foundation award as part of the new AI-ENGAGE program. The project uses AI-driven genomic selection models to develop wheat varieties that produce more food and hold up better against tough growing conditions. It's a collaboration between researchers in the U.S., Japan, and India.
Your takeaway: This is AI working on a problem that affects everyone: feeding more people with less. While most AI headlines focus on chatbots and tech drama, this is an example of the technology being pointed at something we actually need.
The story: ChatGPT uninstalls jumped 295% in a single day after OpenAI announced a partnership with the U.S. Department of Defense. One-star reviews exploded by 775%. Meanwhile, downloads for Anthropic's Claude surged 51% and the app hit #1 on the U.S. App Store for the first time, beating ChatGPT in daily downloads.
Your takeaway: Nothing changed about how ChatGPT works. People deleted it because of political climate, not because of any technical problem. If you rely on any AI tool for work, school, or daily life, this is a good reminder to always have a backup ready.
TOOLS ON OUR RADAR
🐧 Joplin Free and Open Source: A highly secure application that handles massive collections of notes and syncs them seamlessly across all your devices using your own private cloud storage. (Alternative to Evernote)
✂️ *CapCut Freemium: An incredibly versatile video editing application that provides professional grade transitions and effects making it incredibly simple to produce highly engaging content for your social channels.
📸 Photoroom Freemium: A highly intuitive photo editing application that instantly removes backgrounds and generates studio quality lighting for your product images in seconds.
🎵 Udio Freemium: A highly expressive artificial intelligence music creator that produces astonishingly realistic tracks across any genre strictly from your conversational text prompts.
TRENDING
Survey Finds 45% of Cybersecurity Leaders Work a Sixth Day as AI Redefines Their Role - A new survey shows nearly half of cybersecurity leaders are working six-day weeks as AI reshapes their jobs. Security teams are prioritizing data protection, faster threat detection, and figuring out how AI fits into their defenses.
AI Biases Can Influence People's Perception of History - New research finds that AI-generated summaries of historical events can shift how people feel about social and political issues, even when the AI isn't trying to persuade anyone. The biases come baked into the training data.
Scientists Make a Pocket-Sized AI Brain With Help From Monkey Neurons - Researchers shrank an AI vision model from 60 million variables down to just 10,000 by studying how monkey brains process images. The tiny model still works nearly as well, suggesting AI systems could be way smaller and use far less energy.
Waymo Robotaxi Blocks Ambulance Responding to Austin Mass Shooting - Video shows a driverless Waymo taxi stuck in the middle of a street, blocking an ambulance racing to the scene of a deadly shooting that killed three and injured 14. A police officer had to speak through the car's speakers to get it to move.
Ars Technica Fires Reporter After AI Fabricated Quotes in a Story - Ars Technica fired its senior AI reporter after a story included fake quotes generated by ChatGPT and attributed to a real person. The reporter said he was sick and accidentally used AI-paraphrased text instead of actual quotes. The irony: he covered AI risks for a living.
China's Agibot to Make Robots for Auto Industry in Europe - Chinese startup Agibot is partnering with auto supplier Minth Group to manufacture humanoid robots in Europe. The company shipped over 5,000 robots in 2025, while U.S. competitors like Figure AI and Tesla shipped around 150 each.
How Will We Know if AI is Smart Enough to Do Science? - Dozens of new benchmarks are trying to test whether AI can truly reason like a scientist, but researchers can't agree on the right approach. One popular test uses 2,500 expert-level questions, but critics say knowing obscure trivia isn't the same as doing real research.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🎴 Build a flashcard app with create and study modes, card flipping, and mastered-card tracking
Build me a single-file HTML app called "Flashcard Maker" using React 18
(CDN) + Babel standalone. Dark green color scheme (#0a1f0a background,
#50c878 accent). No external CSS or imports.
Include:
- Create mode: front/back textarea inputs, deck name, Add Card button, and
a deck list showing all cards with a "Mastered" checkmark fo
- Study mode: a flippable card (click to reveal answer), Next and Got it
buttons, and a progress panel with total/remaining/mastered + progress bar
- Tab buttons to switch between Create and Study modes
- Search Best Practices button cycling through 4 spaced repetition tips
Use only React.useState, i
Make it fully functional in a single HTML file.What this does:
Builds a complete study tool you can actually use. Create cards with questions and answers, organize them into a named deck, then flip to study mode to test your recall. Mark cards you know as “Got it” and they drop out of your study queue. The progress bar shows mastered vs. remaining at a glance.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Shrink a 60-million-variable vision model down to 10,000 variables by learning from how monkey brains work
❌ Still Can't: Drive a car that knows to get out of an ambulance's way immediately during an emergency
✅ AI Can Now: Help scientists identify wheat varieties that produce more food across three countries
❌ Still Can't: Give you a factual history summary without quietly nudging your political opinions
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Shot entirely in black and white, this episode follows a woman being relentlessly hunted by a robotic 'dog' across a post-apocalyptic landscape. It's basically a 40-minute chase scene, and it's terrifyingly effective. The robot design was inspired by Boston Dynamics, which makes the whole thing hit a little too close to home.
Here's how I use Attio to run my day.
Attio's AI handles my morning prep — surfacing insights from calls, updating records without manual entry, and answering pipeline questions in seconds. No searching, no switching tabs, no manual updates.
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.
