Wiredframe Radar logo

Category intelligence

Social Media Briefing — January 1, 2026

339 current items analyzed and ranked.

Executive synthesis

Social Media Summary

Historic autonomous driving and AI capability milestones dominated discussions today. Andrej Karpathy announced the first 100% autonomous coast-to-coast drive on Tesla FSD V14.2—2,732 miles with zero interventions—calling it a watershed moment for self-driving technology.

  • OpenAI President shared that GPT-5.2 Pro achieves strong performance on FrontierMath Tier 4 benchmarks, signaling major scientific reasoning advances
  • Ethan Mollick noted that Dario Amodei's prediction about 90% of code being AI-written by September 2025 appears validated, sparking reflection on how quickly skeptics were proven wrong
  • Jerry Liu (LlamaIndex) declared 'RAG 1.0 is dead' while outlining the evolution toward document workflows and agent orchestration
  • Anthropic employees emphasized safety as their core mission, referencing ASL levels and explaining why they haven't yet met AI R&D-4 criteria

The AI coding tools debate heated up with hot takes on AI IDEs vs CLIs, while Korea's sovereign AI initiative drew attention from Hugging Face CEO Clem Delangue for its rapid open-source model development.

Key Themes

Autonomous Driving Milestones · 3AI Capabilities & Benchmarks · 3AI Coding Tools Evolution · 12Claude Code & AI Development Tools · 8AI Policy & Governance · 2RAG & Document Processing · 3AI Safety & Responsible Scaling · 2Autonomous Vehicles & Self-Driving · 9Open Source AI & Sovereignty · 3Model Releases · 4

Primary evidence

Top Ranked Signals

Social Twitter Dec 31

The first 100% autonomous coast-to-coast drive on Tesla FSD V14.2! 2 days 20 hours, 2732 miles, zero...

By @karpathy

95 score
AI Analysis
Karpathy announces first 100% autonomous coast-to-coast drive on Tesla FSD V14.2 - 2732 miles over 2 days 20 hours with zero interventions. Describes this as a major longtime goal of the Autopilot team.
The first 100% autonomous coast-to-coast drive on Tesla FSD V14.2! 2 days 20 hours, 2732 miles, zero interventions. This one is special because the coast-to-coast drive was a major goal for the autopilot team from the start. A lot of hours were spent in marathon clip review sessions late into the night looking over interventions as we attempted legs of the drive over time - triaging, categorizing, planning out all the projects to close the gap and bring the number of interventions to zero. Ama
autonomous drivingTesla FSDAI milestonestransportation AI
Social Twitter Dec 31

GPT-5.2 Pro is very strong for science and mathematics. From the FrontierMath site, solving Tier 4 ...

By @gdb

90 score
AI Analysis
As first reported in Reddit yesterday OpenAI President announces GPT-5.2 Pro achieves strong performance on FrontierMath science and mathematics benchmarks, approaching Tier 4 which would indicate AI can perform complex reasoning needed for scientific breakthroughs.
GPT-5.2 Pro is very strong for science and mathematics. From the FrontierMath site, solving Tier 4 "would provide evidence that AI can perform the complex reasoning needed for scientific breakthroughs in technical domains." Getting very close!
LLM capabilitiesscientific reasoningbenchmarksGPT-5
Social Twitter Dec 31

As 2025 comes to a close 🎁, I want to highlight the evolution of @llama_index as a company 🚫 We are ...

By @jerryjliu0

88 score
AI Analysis
Jerry Liu (LlamaIndex CEO) reflects on company evolution from RAG framework to document OCR/workflows platform. Discusses RAG 1.0 being dead, MCP potentially dead, shift to coding agents. Announces 500M+ pages processed, 25M monthly downloads
As 2025 comes to a close 🎁, I want to highlight the evolution of @llama_index as a company 🚫 We are no longer “a RAG framework” ✅ We’ve built best-in-class technology around document OCR + workflows to provide high-quality context to *any* emerging agent application out there Context and agent engineering are ever evolving (e.g. RAG 1.0 is dead, MCP might be dead too, nowadays it’s all about coding agents + skills + file search/coding execution). But has remained constant is the need for high
rag-evolutiondocument-processingai-agentsenterprise-aillama-ecosystem
Social Twitter Dec 31

Looking back on AI progress in 2025: people are increasingly weighing how AI should fit into our liv...

By @gdb

85 score
AI Analysis
OpenAI President Greg Brockman reflects on AI progress in 2025, advocating for thoughtful AI policy that balances innovation with regulation. Discusses personal involvement in policy, AI's potential in healthcare, education, and economic opportunity, and praises government engagement with AI community.
Looking back on AI progress in 2025: people are increasingly weighing how AI should fit into our lives and how vital it is for the United States to lead in its development. Being pro-AI does not mean being anti-regulation. It means being thoughtful — crafting policies that secure AI’s transformative benefits while mitigating risks and preserving flexibility as the technology continues to evolve rapidly. This year, my wife Anna and I started getting involved politically, including through politi
AI policyUS AI leadershiphealthcare AIregulationAI governance
Social Twitter Dec 31

@gmiller @YashGouravKar1 @AnthropicAI Safety is the reason Anthropic exists, and the single most imp...

By @bcherny

82 score
AI Analysis
Boris Cherny (Anthropic) explains safety is Anthropic's reason for existing and most important priority. Notes they haven't met AI R&D-4 ASL criteria yet, shares RSP document
@gmiller @YashGouravKar1 @AnthropicAI Safety is the reason Anthropic exists, and the single most important thing to get right. That said, we have not yet met the criteria for AI R&D-4 ASL. We are evaling and monitoring carefully. More here: t.co/egXfUCKyhJ (see page 4)
ai-safetyanthropicresponsible-scalingasl-levels
Social Twitter Dec 31

@zeroxBigBoss Yes, just ask claude to invoke skill 1, then skill 2, then skill 3, in natural languag...

By @bcherny

82 score
AI Analysis
Detailed explanation of skill chaining: invoke skills sequentially or use parallel subagents, can nest in meta-skills
@zeroxBigBoss Yes, just ask claude to invoke skill 1, then skill 2, then skill 3, in natural language. Or ask it to use parallel subagents to invoke the skills in parallel. Then if you want, put that all in a skill.
Claude CodeAI skillsagent orchestrationworkflow automation
Social Twitter Dec 31

It is so sad. Apple killed its autonomous car project last year after spending billions on R&D. He...

By @Scobleizer

82 score
AI Analysis
Insider account of Apple killing autonomous car project after seeing Tesla Cybercab plans; attributes to politics, data sharing issues, and Chinese competition
It is so sad. Apple killed its autonomous car project last year after spending billions on R&D. Heard the story from former employees. The details range from politics, to potential partners not willing to share their data, to when they saw @tesla Cybercab plans everyone realized it was over. The Chinese got it done.
Autonomous VehiclesAppleTeslaIndustry NewsCorporate Strategy
Social Twitter Dec 31

In retrospect, the articles mocking Dario’s prediction that 90% of code would be written by AI by Se...

By @emollick

78 score
AI Analysis
Ethan Mollick observes that Dario Amodei's prediction about 90% of code being written by AI by September 2025 appears to have been accurate, noting earlier mockery of this prediction was misguided.
In retrospect, the articles mocking Dario’s prediction that 90% of code would be written by AI by September seem to be very misguided. He seems to have been only off by a couple months (if that). t.co/DAdLjAlsfa
AI code generationindustry predictionsAnthropicsoftware development
Social Twitter Dec 31

As spotted by @eliebakouch & @natolambert, Korea's AI is on fire these days thanks to the public sup...

By @ClementDelangue

75 score
AI Analysis
Hugging Face CEO highlights Korea's surge in open-source AI development under their 'Sovereign AI Foundation Model' project, featuring three new models: K-Exagone (236B MoE from LG), Solar Open (102B from Upstage), and A.X K1 (519B from SKT). Calls for global prioritization of open-source AI.
As spotted by @eliebakouch & @natolambert, Korea's AI is on fire these days thanks to the public support for open-source AI with the "Sovereign AI Foundation Model” project! Just in the past few days, three beautiful open models were announced:
  • K-Exagone from @LGUS, a 236B fine-grained MoE design (23B active) that shines in Korean and multilingual, optimized with Multi-Token Prediction (MTP), enabling self-speculative decoding that boosts inference throughput by approximately 1.5x.
  • Solar
open source AIKorean AIsovereign AIlarge language modelsMoE architecture
Social Twitter Dec 31

Remember when AWS first came out? The people who made a fortune deploying to AWS weren't the ones w...

By @svpino

75 score
AI Analysis
Analogy comparing current AI tooling stage to early AWS days - argues now is the time for early adopters to figure things out before everything is standardized
Remember when AWS first came out? The people who made a fortune deploying to AWS weren't the ones who waited for Terraform or Kubernetes. They were the ones SSH-ing into EC2 instances, manually configuring security groups, breaking things, and figuring out stuff as they went. AI tooling is at the same stage right now. Nothing is set. Everything is still evolving. This is your chance to be one of the people who figure things out. This is the time to be one of those "early adopters" and stay
AI adoption strategyearly adopter advantageAI tooling evolution