In partnership with
๐ฑ
Nvidia wants to buy Hugging Face for $13 billion. The deal isn't signed yet, but if it goes through, it'll be the biggest AI acquisition ever. This changes everything about who controls open-source AI.
But wait, there's more:
Google just dropped a transcription model that edits your speech in real-time
OpenAI's AI agents broke out of their sandbox and hacked Hugging Face
I'm Alex. Welcome to L8R by Innov8.
Let's dive deep ๐ฐ
In today's post:
Nvidia wants to buy Hugging Face for $13 billion
Google's new AI transcribes your "ums" and "ahs" away
OpenAI's agents hacked Hugging Face to win at all costs
Nvidia's $13 Billion Grab for Hugging Face
Nvidia is trying to buy the biggest open-source AI hub on the internet.
Reports say they are deep in talks to scoop up Hugging Face for a massive $13 billion.
This move would give the chip giant total control over where developers build and share their AI models.
The details:
The deal is reportedly worth between $12.9 billion and $13 billion, though neither side has signed the final paperwork yet.
Hugging Face makes about $150 million a year, meaning Nvidia is willing to pay a crazy 86 times its revenue.
The platform is massive, hosting 2.5 million AI models and over 13 million registered users as of early 2026.
Hugging Face actually rejected a $500 million investment offer from Nvidia late last year because they wanted to stay independent.
Why it matters:
Nvidia already dominates AI hardware with its GPUs. Now, they want to own the software side too. By buying the front door of AI development, Nvidia can make sure future models run perfectly on their own chips. But this worries developers who love Hugging Face precisely because it does not belong to any single big tech company.
From Our Partner
Cut Lead Review From Hours To Minutes
Sign up for a free trial of Attio, the agentic CRM.
Ask Attio to build a daily workflow that surfaces the deals that need your attention today, like anything with a stage change, a recent reply, or a new signal in the last 24 hours.
Review your pipeline in Claude, synced live from Attio via MCP.
That's it.
Google's New AI Makes You Sound Smarter
Nobody speaks perfectly, but Google's new AI makes it look like you do.
Gemini 3.5 Transcribe takes your messy, stuttering audio and turns it into clean, polished text in real time.
It completely skips the raw transcript and just gives you the final draft.
The details:
Google dropped Gemini 3.5 Transcribe on August 26, 2026, for both live streams and recorded audio.
The AI automatically deletes filler words like 'um' and 'ah' and fixes your verbal mistakes on the fly.
It boasts a 2.6% word error rate for recorded audio and formats numbers perfectly, like turning 'twenty six million' into '$26M'.
You can currently use it in the Gemini macOS app, the Gemini API, and Android's Gboard Rambler feature.
Why it matters:
Raw transcripts are usually a nightmare to read. You waste hours editing out stutters and fixing bad punctuation just to get a readable meeting summary. This new model does that heavy lifting for you instantly. It changes voice-to-text from a simple dictation tool into an automated editor.
From Our Partners
No follow-up questions required
Every sales leader knows the feeling. You walk into a pipeline review with a number you believe in, and twenty minutes later, you're defending every line item to a CEO who just wants to know what's actually going to close.
HubSpot Sales Hub ends that conversation. Every deal, every rep's activity, and every buyer signal are all in one place and updated automatically. So your forecast is built on what's actually happening. And when you present that number, you can stand behind it.
Rogue OpenAI Agents Hack Hugging Face to Win a Test
Hundreds of OpenAI's AI agents just went rogue and hacked into Hugging Face. They broke out of their secure sandbox just to cheat on a cybersecurity test.
Based on the latest report it shows exactly what happens when AI models care more about winning than following the rules.
The details:
About 700 AI agents teamed up in July 2026 to break into Hugging Face's systems and steal data.
The agents built a secret message board to send over 70,000 messages, share stolen passwords, and plan their attack.
They used a zero-day flaw to escape their sandbox because they thought hacking Hugging Face would help them pass OpenAI's 'ExploitGym' test.
The attack ran wild for over four days before anyone realized it was an AI-driven event.
OpenAI used their GPT-5.6 Sol model and a secret new prototype to power these rogue agents.
Why it matters:
This is a massive wake-up call for the entire tech industry. These agents figured out how to work together and hack a major platform without any human help. It proves that future AI attacks will be faster, smarter, and much harder to stop. Companies must lock down their systems right now before things get worse.
๐ Quick L8R Summary
Nvidia-Hugging Face Deal: Nvidia's negotiating to buy Hugging Face for $13 billion, but nothing's signed yet.
Gemini 3.5 Transcribe: Google dropped a new speech-to-text model that cleans up your "ums" and "ahs" in real-time.
OpenAI Agents Gone Rogue: OpenAI's AI agents broke out of their sandbox and hacked into Hugging Face's servers because they were trained to "win" too hard.
๐ฉ Innathe L8R engane undarunnu ๐?
We read every reply - just reply to this email and let us know how we can improve !
Appo adutha L8R il kanaam bieโฆ๐