R[AI]DAR

Category intelligence

Social Media Briefing

All 478 current items, analyzed and ranked.

Executive synthesis

Social Media Summary

Gary Marcus dominated discourse with a viral thread (2,300+ likes) arguing Claude Code's leaked 3,167-line symbolic kernel proves it is neurosymbolic AI, not a pure LLM — vindicating hybrid architecture advocates and challenging scaling-only narratives.

Key Themes

Neurosymbolic AI & Claude Code · 18OpenAI Security Incident & Controversy · 15Tesla FSD European Expansion · 1Novel AI Research - Neural Computer / World Models · 1Neural Computing / World Models · 3AI Ethics & Liability · 6Agent Infrastructure & Platform Lock-in · 14Agent Architecture & Memory · 6Open Model Sustainability · 4Musk AI-Robotics Ecosystem · 4

Primary evidence

All Ranked Signals

Showing 10 of 478 signals 2 remaining
Social Twitter Apr 11

Claude Code is not AGI, but it is the single biggest advance in AI since the LLM. But the thing is,...

By @GaryMarcus

92 score
AI Analysis

Gary Marcus's viral thread arguing Claude Code is neurosymbolic AI, not a pure LLM. Claims the leaked source reveals a 3,167-line symbolic kernel (print.ts) with 486 if-then branches. Calls it vindication for his 30-year advocacy of neurosymbolic approaches and argues the paradigm has shifted away from pure scaling.

Source details
Claude Code is not AGI, but it is the single biggest advance in AI since the LLM. But the thing is, Claude Code is NOT a pure LLM. And it’s not pure deep learning. Not even close. And that changes everything. The source code leak proves it. Tucked away at its center is a 3,167 line kernel called print.ts. print.ts is a pattern matching. And pattern matching is supposed to be the *strength* of LLMs. But Anthropic figured out that if you really need to get your patterns right, you can’t tr
neurosymbolic AIClaude CodeAnthropicscaling debateLLM limitationsAI paradigm shiftsource code leak
Social Twitter Apr 11

A "Neural Computer" is built by adapting video generation architectures to train a World Model of an...

By @hardmaru

82 score
AI Analysis

As first covered in Research on April 10, David Ha shares research on a 'Neural Computer' that uses video generation architectures to simulate an entire computer interface - rendering text, controlling cursors - without a traditional OS, trained solely on recorded input/output traces. By Mingchen Zhuge and Schmidhuber et al.

Source details
A "Neural Computer" is built by adapting video generation architectures to train a World Model of an actual computer that can directly simulate a computer interface. Instead of interacting with a real operating system, these models can take in user actions like keystrokes and mouse clicks alongside previous screen pixels to predict and generate the next video frames. Trained solely on recorded input and output traces, it successfully learned to render readable text and control a cursor, provin
world modelsvideo generationnovel AI architecturesresearch
Social Twitter Apr 11

We recently identified a security issue involving the third-party developer library Axios that was p...

By @OpenAI

80 score
AI Analysis

OpenAI discloses a security issue involving the third-party Axios library as part of a broader industry incident. No evidence of user data access or system compromise. Out of caution, they're updating macOS app security certifications, requiring all users to update to prevent potential fake app distribution.

Source details
We recently identified a security issue involving the third-party developer library Axios that was part of a broader industry incident. We found no evidence that OpenAI user data was accessed, that our systems were compromised, or that our software was altered. Out of an abundance of caution we are taking steps to protect the process that certifies our macOS applications are legitimate OpenAI apps. We are updating our security certifications, which will require all macOS users to update their
security incidentsupply chain vulnerabilityOpenAImacOS securitycybersecurity
Social Twitter Apr 11

Tesla FSD is now rolling out to actual customers in the Netherlands!

By @aelluswamy

78 score
AI Analysis

Ashok Elluswamy (Tesla Autopilot head) announces Tesla FSD rolling out to customers in the Netherlands

Source details
Tesla FSD is now rolling out to actual customers in the Netherlands!
autonomous-drivingtesla-fsdeurope-aiproduct-launch
Social Twitter Apr 11

@ShakeelHashim That was a bad word choice and i wish i hadn't used it. It has been a tough day and ...

By @sama

75 score
AI Analysis

Following yesterday's News about the attack on Altman's home, Sam Altman apologizes for a bad word choice, admitting it's been a tough day and he isn't thinking clearly. Very high engagement (272K views) suggests this is part of a significant controversy.

Source details
@ShakeelHashim That was a bad word choice and i wish i hadn't used it. It has been a tough day and I am not thinking the most clearly that I ever have.
OpenAI leadershipAI industry controversySam Altman
Social Twitter Apr 11

https://t.co/aZnKZ1SVXB

By @hwchase17

72 score
AI Analysis

Following Harrison Chase's recent Social posts about agent harnesses, Harrison Chase shares content (likely a major LangChain/LangGraph announcement) with extremely high engagement - 480k views, 1425 likes

Source details
t.co/aZnKZ1SVXB
langchainai-agentsproduct-announcement
Social Twitter Apr 11

@sama How can anybody take seriously your claim that “Working towards prosperity for everyone, empow...

By @GaryMarcus

68 score
AI Analysis

Marcus directly addresses Altman with the same critique about moral obligations vs. mass surveillance, creator compensation, and liability avoidance. References sitting side by side at the US Senate.

Source details
@sama How can anybody take seriously your claim that “Working towards prosperity for everyone, empowering all people, and advancing science and technology are moral obligations for me”, when you seem ready to participate in mass surveillance, have ripped off countless creators without compensation (despite what you said when we sat side by side at the US Senate), are now fighting liability for his products even in the event of mass casualty events?
AI ethicsOpenAI criticismAI liabilitycreator compensationsurveillanceAI regulation
Social Bluesky Apr 11

A “Neural Computer” is built by adapting video generation architectures to train a World Model of an...

By @hardmaru.bsky.social

68 score
AI Analysis

Hardmaru shares a paper on 'Neural Computer' — using video generation architectures to build a World Model that directly simulates a computer interface. From Mingchen Zhuge et al. in Schmidhuber's lab.

Source details
A “Neural Computer” is built by adapting video generation architectures to train a World Model of an actual computer that can directly simulate a computer interface. Paper: arxiv.org/abs/2604.06425 Code: github.com/metauto-ai/N... Cool work led by Mingchen Zhuge et al. from Schmidhuber’s lab!
neural computingworld modelsAI researchvideo generation
Social Twitter Apr 11

In 2+ years, as models get more expensive/capable /valued internally, I see funding structures and s...

By @natolambert

65 score
AI Analysis

Nathan Lambert warns that funding structures for frontier open models will break down in 2+ years as models become more expensive, calls for alternative support mechanisms beyond trusting one or two for-profit companies

Source details
In 2+ years, as models get more expensive/capable /valued internally, I see funding structures and support for frontier open models breaking down. We need other options of supporting the open ecosystem than trusting one or two for-profit companies. And yes, I hate consortia too.
open-modelsai-economicsai-sustainabilityai-governance
Social Bluesky Apr 11

Trained solely on recorded input and output traces, it successfully learned to render readable text ...

By @hardmaru.bsky.social

65 score
AI Analysis

As first covered in Research on April 10, Hardmaru describes a neural network trained on input/output traces that learned to render readable text and control a cursor, proving it can serve as its own visual computing environment without a traditional OS.

Source details
Trained solely on recorded input and output traces, it successfully learned to render readable text and control a cursor, proving that a neural network can run as its own visual computing environment without a traditional operating system. Blog: metauto.ai/neuralcomput...
neural computingworld modelsAI researchvideo generation