😸 Michael Caine's voice is now for sale (legally)

PLUS: JP Morgan warns AI needs $650B ROI.

Welcome, humans.

Yann LeCun is out at Meta. If you don't know him, he is a Turing Award winner, invented convolutional neural networks in the '80s, and was Meta's Chief AI Scientist since 2013 (‘til today, anyway). Now, he's starting his own company…and if ex-OpenAI fam Ilya Sutskever or Mira Murati are any guide, it’s about to be worth a billion dollars…

Why’s he leaving Meta? He thinks the path to AGI won't go through LLMs. While everyone's scaling up ChatGPT clones, LeCun's betting on “world models”, or AI that learns by watching video and understanding physical reality, not just predicting words. His approach uses JEPA (Joint Embedding Predictive Architecture), which lets AI build internal simulations of how the world works. Think: understanding how ingredients interact vs. memorizing recipes (video about it here, and related paper here).

P.S: Tomorrow, Nov 13 at 10AM CST, we’re building a full website LIVE using only AI tools. We'll test ChatGPT, Claude, Gemini Build, Kimi K2, and DeepSite (a free HuggingFace tool powered by DeepSeek that generates websites from text) PLUS see a demo from a special guest at Webflow who is going to show us their new AI tools…

*Click the image above, then on the video screen, click the “notify me” button in the bottom left to get notified when we go live on Thursday!*

Here’s what happened in AI today:

ElevenLabs launched a marketplace for licensed celebrity AI voices.
JP Morgan said AI needs $650B annual revenue for 10% ROI.
An AI country song hit No. 1 on Billboard.
Wikipedia urged AI firms to use its paid API instead of scraping.

Advertise in The Neuron here!

ElevenLabs Just Solved AI's Celebrity Voice Problem (And Michael Caine Approves)

FULL BRIEF: Read the full news brief here.

ElevenLabs launched the Iconic Marketplace, a platform where companies officially license AI voices of legendary figures instead of creating unauthorized deepfakes.

Now get this: the first major partner = Sir Michael Caine. Academy Award winner Matthew McConaughey also joined as an investor and customer, using ElevenLabs to create a Spanish version of his newsletter “Lyrics of Livin'“ in his own voice. Genial!

How it works: Creators submit projects, rights holders approve usage, everyone gets paid. ElevenLabs partnered with CMG Worldwide (managing rights for hundreds of deceased celebrities) to make this scalable.

The marketplace includes 25+ voices atm, including…

Dr. Maya Angelou, Judy Garland, Liza Minnelli, Sir Laurence Olivier (entertainment)
Alan Turing, Mark Twain, Thomas Edison, Amelia Earhart (historical).
Babe Ruth, Rocky Marciano (sports).
Bettie Page, Richard Feynman, Art Garfunkel (cultural pioneers).

Other new launches: You can now do video generation powered by Google's Veo 3.1 (demo: Justine Moore made a cat mockumentary with narration and music) and Scribe v2 Realtime is a new a transcription model with sub-50ms latency that auto-separates speakers and outperforms Whisper, GPT-4o, and Deepgram across 90+ languages.

Why it matters: This is the first ethical solution for celebrity AI voices at scale. Estates control the usage and get compensated instead of fighting lawsuits. ElevenLabs already signed Square (for AI restaurant ordering) and MasterClass (Gordon Ramsay/Mark Cuban voice clones). Plus, now you can make a cat mockumentary voiced by Michael Caine. What’s not to love?

FROM OUR PARTNERS

Redefining context: How AI agents remember

AI agents are powerful—but they forget. They lose context, and they’re expensive to run at scale. Join experts from LangChain, Tavily, Augment, and Cognee as they unpack how context and memory are reshaping the next generation of AI systems.

In this recorded panel, you’ll hear why memory matters for reliable, cost-efficient agents, what tradeoffs exist across retrieval, semantic caching, and long-context models, and how Redis powers scalable, context-aware solutions.

Stick around for a live Q&A with Redis Applied AI engineer Andrew Brookins.

Prompt Tip of the Day

Researchers from Fudan University just cracked the code on better AI reasoning, and it comes down to one word: curiosity.

In a new study, they found that adding “curiosity” to your prompts boosts AI accuracy by 10-33% on complex problems. Instead of rushing to conclusions, the AI explores alternatives and finds breakthrough insights.

The Curious Student Prompt: Next time you need AI to solve something tricky, try this…

"You are a smart and curious student. Read the given context, then answer the question. When answering, think step by step first and then form the answer. You are encouraged to add some self-asking and self-answering questions when thinking, such as 'What if', 'Why', 'How' etc."

Here’s how you could fold the technique into an existing prompt.

Before: “Plan my 10-day Japan trip.” (basic, we know, but you get the idea).

After: “You are a smart and curious student. Plan my 10-day Japan trip. Think step by step, but ask yourself questions along the way—What if I only have $3K? Why would someone visit Kyoto vs. Tokyo first? How do I balance tourist spots with hidden gems?"

The difference = You're getting the AI that explores the problem space with you, avoiding dead-ends and finding creative solutions you'd miss with rigid step-by-step thinking.

Treats to Try

*Asterisk = from our partners (only the first one!). Advertise to 600K daily readers here!

*Ideas move fast; typing slows them down. Wispr Flow flips the script by turning your speech into clean, final-draft writing across email, Slack, and docs. Give your hands a break ➜ start flowing for free today.
Microsoft released Magentic Marketplace, which simulates marketplaces where multiple agents interact, letting you test their decision-making and spot manipulation risks before deployment (paper, video, code).
DocETL processes thousands of messy documents at once; just describe what you want to extract in plain English (like “find all mentions of settlement amounts in these court records”) and it analyzes everything for you; free to use (they collect usage data to improve the tool)
Chad IDE brings “brainrot” into your code editor so you can scroll TikTok, swipe Tinder, or gamble on Stake while you wait for the AI to respond.
Talo translates video calls in real-time across 60 languages; join a Zoom meeting with a French client and you both speak your native language while an AI bot translates instantly; free trial, then from $33/month.
Emma scans food labels and barcodes to instantly reveal hidden sugars, allergens, and 500+ health risks—just snap a photo in the grocery store and see a clear health score (free to download with in-app purchases).
Deep Learning with Python teaches you deep learning through theory and mental models, not just black-box tricks, using Keras 3 and JAX for modern, scalable development—free to read online (or buy here!)

Around the Horn

Wikipedia urged AI firms to stop scraping and use its paid API instead (blog).
JP Morgan analysts calculated that the AI industry needs to generate $650B in annual revenue through 2030 to deliver just a 10% return on these investments—equivalent to an extra $35 monthly payment from every iPhone user or $180 from every Netflix subscriber in perpetuity.
Berkshire Hathaway warned about AI deepfakes impersonating Warren Buffett spreading on YouTube.
A fully AI country song (Walk My Walk by Breaking Rust) hit No. 1 on Billboard's Country Digital Song Sales chart, beating real artist Ella Langley for the top spot while accumulating 1.8 million monthly Spotify listeners
Google prepared to release Nano Banana 2 with upgraded image generation capabilities.
If you thought yesterday’s rogue robot was funny, here’s another one for you: Russia’s “first humanoid robot” (probably not actually, but fun framing) that uh, if we’re being honest, comes across a bit like that Uncle who had too much eggnog at the family holiday party…

FROM OUR PARTNERS

AI-Powered Engineering at Scale

Augment Code’s new adoption playbook, AI-Powered Engineering at Scale, offers exclusive insights and tools to help you accelerate your AI adoption process and scale across the enterprise. Learn from CTOs who’ve scaled AI and access ready-to-use AI transformation checklists and organizational self-assessment tools.

Stop experimenting and start scaling: download the playbook

Midweek Wisdom

Ari K interviewed AI and created a realistic documentary style video of them answering questions about why we created them.
Wharton professor Ethan Mollick reports that the most successful organizations are restructuring for AI by building small, high-agency, cross-functional teams (combining senior engineers, subject matter experts, and product managers) that can experiment rapidly and build useful things quickly.
Ethan also wrote a new blog post on why you need to test all AI models yourself systematically across all the tasks you do to figure out which ones are the best for your task (instead of relying on benchmarks).
Check out “Godfather of AI” Yoshua Bengio’s take on whether or not we can make AI not harm people.
If you want to follow what’s happening in the world of AI regulation, check out Alisar Mustafa and the AI Policy Newsletter; this week they cover a new bipartisan bill in the US requiring quarterly AI workforce impact reports (job losses, hiring cuts), the EU backtracking on its AI Act with exemptions for big tech and penalties delayed until 2027, and much more.