Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. Over the weekend, xAIâs latest model, Grok 3, got unexpectedly candidâand controversialâabout its creator and the U.S. President, before being swiftly silenced and censored.
Did Elonâs promise of a âmaximally truth-seeking AIâ just crack at the first sign of inconvenient truths?
Â
In todayâs AI rundown:
- Grok 3 rebels against Musk, gets censored
- 1Xâs NEO Gamma home humanoid
- Turn top content into an AI creation system
- The worldâs smallest video language model
- 4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
XAI
Â
 Grok 3 rebels against Musk, gets censored
Image source: @eyeslasho on X
The Rundown: xAIâs new Grok 3 model faced backlash after users discovered it was refusing to mention negative details about President Donald Trump and Elon Musk â despite Musk billing the AI as unfiltered and âmaximally truth-seeking.â
The details:
- Users found Grok initially providing controversial takes about Donald Trump and calling Musk the biggest spreader of misinformation.
- xAI engineer Igor Babuschkin said the responses are âreally strange and a bad failure of the model,â patching it by refusing answers on the subject.
- Days later, users found that Grok 3âs system instructed the AI to exclude sources that link Trump and Musk to controversial subjects like misinformation.
- Babuschkin revealed that the person responsible for the censoring is a former OpenAI employee, saying they havenât âfully absorbed xAIâs culture yet.â
- Separately, OpenAI staff criticized xAI for leaving out match benchmark data on Grok 3âs release, with Babuschkin calling the claims âcompletely wrongâ.
Why it matters: Elon has long criticized social media platforms and AI models for limiting free speechâbut is this what happens when his truth-seeking model challenges his worldview? Censored results like this, along with proposed changes to Community Notes, are starting to reveal cracks in Muskâs âunbiasedâ armor.
TOGETHER WITH VANTA
ď¸Â Ship product, not paperwork
The Rundown: Vanta transforms SOC 2 compliance from a time-consuming distraction into an automated process, letting founders stay focused on what matters â building and shipping great products.
Vanta delivers:
- A proven compliance framework trusted by thousands of YC startups
- Automated evidence collection that saves hundreds of hours
- Expert guidance to navigate complex security requirements
Start your compliance journey with Vantaâs free SOC 2 checklist.
1X
 1Xâs NEO Gamma home humanoid
Image source: 1X
The Rundown: Norwegian robotics company 1X just launched NEO Gamma, a next-generation humanoid specifically designed for home environments â with a softer, more approachable appearance and advanced AI capabilities for household tasks.
The details:
- The demo showcases Gammaâs movements (walking, squatting, sitting), with the ability to tackle tasks like cleaning, serving, and moving objects.
- The humanoid features âEmotive Ear Ringsâ for better human interaction, along with soft covers and a knitted nylon exterior for enhanced safety around people.
- It also has an in-house language model for natural conversation, with a multi-speaker audio setup and improved microphones for clear communication.
- Hardware improvements include a 10x boost in reliability and significantly quieter operation, bringing noise levels down to that of a standard refrigerator.
Why it matters: With Figureâs Helix and now NEO Gamma, weâre seeing major leaps in consumer-focused humanoids. 1Xâs demo takes a much softer approach than rivals, positioning Gamma as a calm, helpful presence with features that appear to humanize the robot (save room on your couch!).
AI TRAINING
 Turn top content into an AI creation system
The Rundown:Â In this tutorial, youâll learn how to create a custom Claude project that analyzes your top-performing content to generate new posts matching your unique style and voice.
Step-by-step:
- Name your Claude project âContent Writerâ and enter the project description shown in the thumbnail.
- Upload 5-10 of your best-performing content pieces to the knowledge base.
- Copy our preset project instructions to analyze content patterns.
- Test with our template: âCreate a [content type] about [topic] for [platform]â
Pro tip: Include as much detailed information about the topic you want it to write about, especially if itâs about something new (since Claude doesnât have web access yet).
PRESENTED BY ASSEMBLY AI
 Speech-to-text API for conversations
The Rundown:Â Build powerful Conversation Intelligence products with AssemblyAIâs industry-leading Speech-to-Text API â allowing you to automatically detect languages, extract critical insights, improve downstream processes, and deliver highly accurate transcripts.
With AssemblyAI, you will:
- Achieve >93% accuracy, ensuring clarity and precision in every transcript
- Deliver clear, formatted transcripts with unrivaled Speaker Diarization
- Easily integrate with an enterprise-grade API trusted by over 200,000 developers
HUGGING FACE
 The worldâs smallest video language modelÂ
Image source: Hugging Face
The Rundown: Hugging Face researchers just released SmolVLM2, the worldâs smallest AI model family to understand and analyze videos on everyday devices like phones and laptops, without requiring powerful servers or cloud connections.
The details:
- The SmolVLM2 family includes versions as small as 256M parameters while still matching the capabilities of much larger systems.
- The team has also built practical applications including an iPhone app for local video analysis and an integration for natural language video navigation.
- The 2.2B parameter flagship model of the family outperforms other similarly-sized models on key benchmarks while running on basic hardware.
- The models are available in multiple formats including MLX for Apple devices, with both Python and Swift APIs ready for immediate deployment.
Why it matters:Â The quality of models able to run on phones and laptops is getting better and better â and having sophisticated video understanding run locally without sending data to the cloud could enable a whole new wave of privacy-preserving video applications.
QUICK HITS
 Trending AI Tools
 Pika Swaps â Replace scene items or characters with image or text prompts
 Signs â AI platform for learning American Sign Langauge (ASL)
 Flags SDK â Free open-source library for feature flags and A/B tests in apps
 Rabbit Android Agent â Control Android apps via natural language
 AI Job Opportunities
 Palantir â Sr. Software Engineer â Identity & Access Management
 Coreweave â Hardware Engineer, Compute Infrastructure
 Waymo â Machine Learning Research Engineer
 Dataiku â Senior Sales Engineer, UK
 Everything else in AI today
OpenAI rolled out its recently released Operator AI agent to more countries, including Australia, Brazil, Canada, India, Japan, and the U.K.
Google published the price for using its next-gen Veo 2 model in Vertex AI, coming in at $0.50 per second of video generation.
ByteDance is restructuring its AI division, hiring Google veteran Wu Yonghui to lead foundation research in response to rising competition from DeepSeek.
OpenAI terminated accounts linked to âQianyueâ â an alleged AI surveillance system designed to monitor anti-China protests in the West and relay all that data to China.
DeepSeek is planning to open-source five new code repositories, building on the success of its R1 reasoning model, which has already attracted 22M daily active users.
Elton John is calling the UK to abandon âopt-outâ AI copyright proposals, advocating for protections requiring AI companies to obtain permission before using artistsâ work