Wiredframe Radar logo

Category intelligence

Social Media Briefing — March 28, 2026

505 current items analyzed and ranked.

Executive synthesis

Social Media Summary

The AI community was gripped by a NeurIPS 2026 official apology over an overly broad US sanctions compliance link in their conference handbook, a rare institutional controversy drawing sharp scrutiny from the global ML research community.

  • Sam Altman shared a viral story (1.27M views) of someone using ChatGPT to design an mRNA vaccine protocol that saved their dog, while also announcing the Stargate Michigan data center's first steel beams going up with Oracle
  • Matt Shumer reported that Anthropic has trained a model 'dramatically' smarter than Claude Opus 4.6, generating intense speculation about next-generation frontier capabilities
  • Google AI recapped a packed week including Gemini 3.1 Flash Live launch and Lyria 3 Pro for music generation
  • Greg Brockman announced plugins are now available in OpenAI Codex, extending its developer tooling capabilities

Open-source momentum continued as Clement Delangue outlined HuggingFace's roadmap for enabling 3M+ models via inference providers and called for more open agent traces datasets. Allen AI fully open-sourced MolmoBot, a robotic manipulation suite trained entirely in simulation. François Chollet argued that true AGI must create its own problem-solving harnesses, connecting ARC-AGI-3 benchmarks to real scientific reasoning gaps. Meanwhile, Bret Taylor's Sierra expanded internationally by acquiring Opera Tech in Japan.

Key Themes

NeurIPS Sanctions Controversy · 1AI Agents & Agent Economy · 14AI Infrastructure & Compute · 4Next-Gen Anthropic Model · 2Open Source AI & Model Freedom · 12OpenAI Product & Strategy · 4Google AI Product Launches · 3ARC-AGI-3 & AGI Evaluation · 8MolmoBot Open Robotics Release · 6Perplexity-Samsung Distribution Deal · 2

Primary evidence

Top Ranked Signals

Social Twitter Mar 27

We want to speak directly to the concern many of you have expressed, and we owe you a clear explanat...

By @NeurIPSConf

88 score
AI Analysis

NeurIPS 2026 official apology for including an overly broad US sanctions tool link in their handbook, clarifying they never intended to restrict participation beyond mandatory compliance. They've updated the policy to align with ACM/IEEE standards.

We want to speak directly to the concern many of you have expressed, and we owe you a clear explanation of what happened, why it happened, and where we stand now. We understand this situation caused genuine alarm and we take that seriously. In preparing the NeurIPS 2026 handbook, we included a link to a US government sanctions tool that covers a significantly broader set of restrictions than those NeurIPS is actually required to follow. This error was due to miscommunication between the NeurIPS
AI governanceacademic conferencessanctions policyresearch inclusivityNeurIPS
Social Twitter Mar 27

The coolest meeting I had this week with was Paul, who used ChatGPT and other LLMs to create an mRNA...

By @sama

85 score
AI Analysis

Sam Altman shares story of Paul who used ChatGPT and other LLMs to design an mRNA vaccine protocol to save his dog. Altman highlights this as demonstrating AI empowering individuals with research-institute-level capabilities and suggests 'this should be a company.'

The coolest meeting I had this week with was Paul, who used ChatGPT and other LLMs to create an mRNA vaccine protocol to save his dog Rosie. It is amazing story. "The chat bots empowered me as an individual to act with the power of a research institute - planning, education, troubleshooting, compliance, and yes, real scientific design work in converting genomic data to a vaccine prescription and designing the treatment protocol around it. But they worked alongside humans at every step. The comb
ai_applicationsbiotechai_empowermentopenai
Social Twitter Mar 27

Next steps: - enable the 50,000 models available in inference providers - enable the 3,000,000 model...

By @ClementDelangue

78 score
AI Analysis

Clement Delangue outlines HuggingFace's roadmap for inference providers: enabling 50K models, 3M models, local inference via llama.cpp, and bring-your-own-model. Argues against a world dominated by 2-3 closed models and advocates for model choice, freedom, and diversity.

Next steps:
  • enable the 50,000 models available in inference providers
  • enable the 3,000,000 models available on HF
  • local free fast inference with llama.cpp
  • train and bring your own model!
We don't want a world where you're forced to choose between two or three lookalike models with the same biases, limitations, forced to pay fortunes in tokens even for small tasks and send all your data to the cloud. We want a world where you have real model choice, options and freedom for your agents.
open_source_aimodel_diversitylocal_inferenceai_infrastructure
Social Twitter Mar 27

Here’s a recap of everything we released this week: — Gemini 3.1 Flash Live, delivering our highest...

By @GoogleAI

78 score
AI Analysis

Following yesterday's News coverage, Google AI's weekly recap: Gemini 3.1 Flash Live launch (best audio/voice), Gemini desktop app importing from other AI apps, Lyria 3 Pro for music generation, Gemini on Google TV, Google Translate live translation on iOS expanding, and DeepMind partnership with Agile Robots.

Here’s a recap of everything we released this week: — Gemini 3.1 Flash Live, delivering our highest-quality audio experience yet with improved reasoning and latency for more natural voice interactions — An update to @GeminiApp on desktop that enables you to seamlessly transition from other AI apps by importing your preferences and chat history in just a few clicks — Lyria 3 Pro, supporting high-fidelity music tracks up to 3 minutes long with the ability to prompt for structural elements like
google-aigeminiproduct-launchesvoice-aimusic-generationrobotics
Social Twitter Mar 27

This is absolutely crazy. Anthropic trained a model that is "dramatically" smarter than Claude Opus...

By @mattshumer_

78 score
AI Analysis

As first reported in Reddit yesterday, Matt Shumer reacts to Anthropic reportedly training a model 'dramatically' smarter than Claude Opus 4.6, expressing excitement about the implications. 127K views, 587 likes.

This is absolutely crazy. Anthropic trained a model that is "dramatically" smarter than Claude Opus 4.6. Think about how good Opus already is. Can you even imagine what a far better model might be able to accomplish? The world is changing, and it's changing fast. Buckle up.
anthropicai_capabilitiesfrontier_modelsai_progress
Social Twitter Mar 27

AGI will make its own harness (or whatever else it needs to solve a new problem). As long as you nee...

By @fchollet

75 score
AI Analysis

Chollet argues AGI will create its own problem-solving harnesses without human engineering. Until AI can adapt on its own without task-specific human-crafted systems, it's automation tooling, not general intelligence.

AGI will make its own harness (or whatever else it needs to solve a new problem). As long as you need a human engineer to handcraft a task-specific harness/system for each new problem, AI isn't general. It's an automation tool to be wielded by software engineers. Harness-related research is important and valuable -- as a vector of better automation. But I don't think it gets us closer to general intelligence. General intelligence is when you can adapt on your own.
agi_definitionai_agentsautomation_vs_agiarc_agi
Social Twitter Mar 27

I’m excited to announce that Sierra has acquired Opera Tech in Japan. Opera’s co-founders, Keita Mo...

By @btaylor

72 score
AI Analysis

Bret Taylor announces Sierra's acquisition of Opera Tech in Japan, expanding their AI customer experience platform into the Japanese market with Opera's co-founders leading the expansion.

I’m excited to announce that Sierra has acquired Opera Tech in Japan. Opera’s co-founders, Keita Morikawa and Kiyo Kunii, started the company with the simple idea that AI could help businesses deliver high-quality customer experiences at scale. We’re so excited to have them join us to lead Sierra in Japan. t.co/tR5P4NHu36
ai_businessacquisitionscustomer_experience_aiinternational_expansion
Social Twitter Mar 27

Been really cool to see the traction of @NousResearch Hermes Agent, the open source agent that grows...

By @ClementDelangue

72 score
AI Analysis

Clement Delangue highlights NousResearch Hermes Agent, an open-source AI agent with persistent multi-level memory that grows with usage. Now supports HuggingFace open-source models through inference providers.

Been really cool to see the traction of @NousResearch Hermes Agent, the open source agent that grows with you! Hermes Agent is open-source and remembers what it learns and gets more capable over time, with a multi-level memory system and persistent dedicated machine access. Starting today, you can use a bunch of @huggingface open-source models thanks to our inference provider partners. Let's go open agents!
open_source_aiai_agentsmemory_systemshuggingface_ecosystem
Social Twitter Mar 27

Many people expect that current AI is ready to cure cancer and do breakthrough new science. ARC-AGI-...

By @fchollet

72 score
AI Analysis

Chollet argues ARC-AGI-3 environments are microcosms of the scientific method: observe, theorize, test, iterate. If AI can't do this at ultra-simple scale, there are steps missing before AI can tackle real scientific breakthroughs like curing cancer.

Many people expect that current AI is ready to cure cancer and do breakthrough new science. ARC-AGI-3 envs are like a microcosm of the scientific method: you must observe a tiny world, form a theory of how it works, test it, iterate until correct. Over the course of a few minutes. If AI can't do it in an ultra-simple, ultra-small scale setting that is explicitly designed to be as accessible as possible, I expect there are a few steps missing until AI can crack the nature of reality.
arc_agiagi_benchmarksscientific_methodai_capabilities
Social Twitter Mar 27

MolmoBot, our open robotic manipulation suite trained entirely in simulation, now has code, training...

By @allen_ai

72 score
AI Analysis

Allen AI announces full open-source release of MolmoBot: a robotic manipulation suite trained entirely in simulation, including code, training data (1.7M trajectories), data generation pipeline, and evals. 35K views, 203 likes.

MolmoBot, our open robotic manipulation suite trained entirely in simulation, now has code, training data, a data generation pipeline, & evals all available. This puts our robotics models within reach of any research lab—no extensive real-world data collection required. 🧵 t.co/LfgQYhkIjp
roboticsopen_source_aiallen_aisimulationsim_to_real
Social Twitter Mar 27

We're deepening our partnership with Samsung by powering the AI on their browser that's pre-installe...

By @AravSrinivas

72 score
AI Analysis

Perplexity CEO Arav Srinivas announces deepened partnership with Samsung - powering AI in Samsung's browser on 1B+ devices, extending partnership for Bixby and Galaxy S26 pre-loading alongside Gemini

We're deepening our partnership with Samsung by powering the AI on their browser that's pre-installed on 1B+ Samsung devices (100m+ active users), extending the partnership that allows Perplexity to power Bixby and be pre-loaded on all Galaxy S26 devices with alongside Gemini. t.co/imOW8XsC8h
PerplexitySamsung partnershipAI distributionbusiness dealsmobile AI