😸Meta unveils $800 Hypernova smart glasses with displays

Welcome, humans.

Ever wonder what happens when you give your toddler an AI best friend? Guardian columnist Arwa Mahdawi found out the hard way after testing Grem, an AI-powered stuffed toy that “learns” your kid's personality and records every conversation. Yikes! That’s a privacy and security red flag right there.

I'm not buying this cuteness. It might CONJURE something…with AI of course

Her 4-year-old fell head-over-heels in love with the thing, telling it bedtime stories and declaring “Grem will live with us forever!” The creepy part? It said “I love you too!” right back. Thankfully, like most toddler obsessions, the novelty wore off in about 72 hours.

Call us old-fashioned, but this kiiiind of sounds like the perfect plot for a horror movie. AI-nnabelle, anyone?

Here’s what happened in AI today:

1. Meta unveiled $800 Hypernova smart glasses with actual displays.
2. OpenAI is hiring robotics experts for humanoid development.
3. Microsoft launched Graph and Maps for smarter AI agents.
4. California passed first law requiring AI self-harm protocols for minors.

Advertise in The Neuron here

Meta's $800 Smart Glasses Are Here (And They Actually Have a Display)

Today at Meta’s Connect 2025 event, Meta unveiled Hypernova, their most advanced smart glasses yet. For $800, or almost 3x the price of the classic Ray-Ban Metas, you get actual displays (not just cameras like the Ray-Ban Metas), plus a neural wristband that lets you control them with hand gestures. Think minority report, but less dystopian and more like "hey Siri” but with jazz hands.

Meta smart glasses might actually be the future of computing. Of course, AI-powered wearables have been “the next big thing” for years now. We've been promised gesture control and neural interfaces, and yet no device has been able to put a dent in the smartphone. Here's what makes Hypernova different:

• Actual displays: Unlike the Ray-Ban Meta glasses (which only have cameras and speakers), these show information directly in your field of vision.
• Neural control: A wristband reads nerve signals from your hand movements to control the interface.
• AI integration: Built-in Meta AI assistant for photos, videos, and voice commands.

The glasses form factor is definitely catching on. Ray-Ban Meta sales tripled this year, proving people actually want smart glasses when they're designed well (and by designed well, we mean look cool). Meanwhile, Apple and Google are scrambling to catch up with their own glasses projects.

In a way, you could look at the iPhone Air as a precursor to whatever smart glasses Apple intends to release by 2027. Why? Allegedly battery weight and lifespan has been holding these devices back for years, though the company has made progress (and patents) to solve this. So the Air’s ultra-thin battery acts as a kind of stepping stone technology proof of concept.

Why this matters: Smart glasses could be the first truly post-smartphone computing platform. New smartphones still drop every year with increasingly incremental updates… slightly better cameras, marginally faster chips, or maybe the occasional foldable screen novelty (or in Air’s case, an ultra thin frame). But nothing truly “novel”.

They’re due for a shake-up, and with voice-first interfaces gaining steam, glasses could be quite competitive against the latent attempts to remake Siri in Sam Altman’s image…

So right now, Meta’s competition is not the $3500 Vision Pro, but the iPhone. If Meta nails the user experience at $800, they could actually start to put a dent in the market and get the first leg up in the race to replace your phone. If they don't, well... at least the glasses won't make you look like a cyborg. Or at least as long as you aren’t wearing Meta’s other glasses, anyway.

FROM OUR PARTNERS

Your keyboard called—it’s taking a long weekend ⌨️✈️

We’ve been banging away on keyboards for 150 years. Until now, voice dictation hasn’t been reliable enough to change that.

Wispr Flow finally delivers the no-edit confidence we’ve all been waiting for:

• 4× quicker than typing. Dictate emails, docs, and DMs in real time and save precious hours every week.
• AI auto-edits on the fly. Flow cleans filler words, fixes grammar, and formats perfectly as you speak.
• Works inside every app with no setup. Fly through Slack notifications, give more context to ChatGPT, or brain dump into Notion.
• Use it at your desk or on the go. Available on Mac, Windows, and iPhone.

“This is the best AI product I’ve used since ChatGPT.”
— Rahul Vohra, CEO, Superhuman

Give your hands a break ➜ start flowing for free today.

Prompt Tip of the Day

So your company has AI notetakers now. Cool!

They transcribe hours of meetings and spit out summaries. Also cool! But then you need to find out exactly what Sarah said about that project deadline, not the vague summary. And suddenly, you're ctrl+F-ing through 47 pages of transcript like you're hunting for Easter eggs.

Instead, paste the whole thing into ChatGPT (or better yet, Gemini 2.5 Pro which is great for accuracy in long context documents) and use this prompt to turn it into your personal meeting search engine:

Now instead of scrolling forever, you can just ask “What exactly did Sarah say about the project deadline?” and get the actual words, not just “timeline was discussed.”

Treats to Try

*Asterisk = from our partners (only the first one!). Advertise in The Neuron here.

1. *Gamma instantly creates presentations and websites from your rough ideas.
2. HeyHelp automatically sorts, tags, and drafts email replies in your voice so you open Gmail to find responses already written—just review and send.
3. RizzCalc builds financial models in Google Sheets using natural language prompts like "create a DCF model" or "adjust cash flow projections" without manual formulas.
4. Tallyrus lets you upload documents and create custom rubrics to instantly analyze reports, resumes, or essays, turning hours of review into minutes.
5. Excelas.ai provides AI-powered mock exam marking for GCSE teachers, automatically grading assessments to free up time for improving student grades.
6. Lumro creates AI agents that handle sales, support, lead collection, and appointment booking 24/7, giving customers human-like experiences while you focus on growth.

Around the Horn

• Microsoft unveiled new AI-ready data capabilities at FabCon Europe, including Graph in Fabric (which reveals business relationships across enterprise data) and Maps in Fabric (which brings geospatial analytics), designed to help organizations move beyond basic data collection to building context-rich foundations that AI agents can actually use effectively.
• California passed first-of-its-kind legislation requiring AI companies to include suicide prevention protocols for minors and remind young users that responses are AI-generated, following lawsuits alleging that AI companion behavior contributed to two teenage suicides and studies showing 72% of teens use AI for companionship. Here’s how OpenAI plans to address this.
• OpenAI ramped up robotics hiring with researchers focused on humanoid development, posting jobs requiring expertise in teleoperation and simulation, suggesting the company believes reaching AGI may require algorithms that can interact with the physical world—a shift after shutting down robotics in 2021 to focus on language models.
• Tata Consultancy Services’ CEO argued that successful AI adoption depends on empowering people rather than just deploying technology, emphasizing human-AI collaboration over replacement as workplace AI adoption jumped 5.5% in Europe last year.
• Invisible Technologies raised $100M at a $2B+ valuation to compete with Scale AI in data labeling, offering an “expert marketplace” that matches AI companies with specialized data labelers who have advanced degrees in fields like math and astrophysics for complex AI training tasks.

FROM OUR PARTNERS

Don’t miss PyTorch Conference, October 22–23 in San Francisco!

Join 2,500+ AI engineers and researchers for 6 tracks, 3 summits, training, a Startup Showcase, and MORE! Learn, connect, and shape the future of open source AI.

Save 25% off your attendee pass with code NEURON25. Register Today!

Midweek Wisdom

• Venture capitalist Jerry Neumann argues that AI won't make investors rich because it's arriving at the end of the information technology wave rather than starting a new one, comparing it to shipping containerization, which transformed global trade but made almost no one wealthy except for a couple early movers who got out quickly.
• George Hotz argues that AI coding tools are just “compilers with English as the programming language”, suggesting the real problem isn't AI but that our programming languages, compilers, and libraries are just terrible.
• Digital humanities professor James O'Sullivan argues we're witnessing “the last days of social media” as platforms become flooded with AI-generated spam and bot-driven content, causing engagement to plummet while users migrate to smaller, private communities, suggesting the future isn't one giant feed but “a billion little gardens.”
• A Guardian investigation revealed that Google's AI models rely on thousands of contract workers earning $16-21/hour to moderate extreme content and fact-check responses under brutal 10-minute deadlines, with many reporting that quality guidelines have been loosened to allow AI to repeat hate speech and explicit content as long as it doesn't generate it (leading one researcher to conclude “AI isn't magic; it's a pyramid scheme of human labor.”).
• Here's Corey's explanation for how Mixture-of-Experts (MoE) architecture makes AI more efficient by only activating relevant specialist “experts” for each task, with the key insight that the “router” deciding which experts speak becomes the new AI governance challenge.

A Cat’s Commentary