Coverage
12 Oct 2026
Super Intelligence (SI) news, papers, lab posts and video briefings from 12 Oct 2026, newest first. Every item links to its source.
Articles
FTAV’s further read
Capability · Financial Times
What's Dragging Down AI Efficiency? The 'Verification Tax.'
Capability · The New York Times
As AI Closed in on 'Unique Games' Proof, Researchers Raced to Beat the Machines
Capability · Quanta Magazine
UK ministers explore compulsory AI safety reporting
Policy · Financial Times
The huge lurking leverage of AI hyperscalers
Capability · Financial Times
Why My Conversations with OpenAI’s ‘ChatGPT for Teens’ Made Me Very Worried
Labs · The New York Times
AI is less regulated than milk, says Microsoft president
Policy · BBC News
OpenAI fires 3 safety researchers in dispute over AI risks
Safety · AP News
Silicon Valley is falling in line with Trump’s attempt to rename AI
Policy · The Verge
Can the AI industry persuade data center opponents by getting rid of NDAs?
Markets · TechCrunch
These execs think voice AI hasn’t reached its ChatGPT moment yet
Labs · TechCrunch
Examining the Doctor-A.I. Relationship
Capability · The New York Times
Learning to use local AI is exciting, overwhelming, and frustrating
Capability · The Verge
The problem with your AI therapist
Capability · Financial Times
What if AI prefers CVs written by AI?
Capability · Financial Times
Academic publishing is a study in fighting AI slop
Capability · Financial Times
Microsoft’s Satya Nadella says AI models need an ‘emergency brake’
Labs · TechCrunch
What are they saying about AI at San Francisco Tech Week?
Capability · NPR
Black Box: The Chatbots | The Line | Ep 5 - podcast
Capability · The Guardian
DistroKid has been quietly taking down songs in response to UMG lawsuit
Capability · The Verge
Nvidia in talks to acquire US ‘open’ model start-up Reflection AI
Markets · Financial Times
Here are the top AI agents that can live in your text messages
Capability · TechCrunch
AI agent makers are promising privacy, will they deliver?
Capability · The Verge
AI Is Getting Really Good at Messing With Cybercriminals
Capability · Wired
UK must not be beholden to foreign AI, says head of Alan Turing Institute
Capability · The Guardian
Jon Davis on job titles in the age of AI - cartoon
Capability · The Guardian
The world of one trade, AI
Capability · Financial Times
Inside the ‘robot gyms’ training machines for the real world
Capability · Financial Times
Why OpenAI’s revenue numbers really matter
Markets · Financial Times
AI surveillance startup Flock to cut several hundred jobs amid backlash, sources say
Markets · The Guardian
Video briefings
AI Is Recreating Photoshop and Other Adobe Apps
Matthew Berman · 74.9K views
Increasingly seeing the stock market is one big bet on the success of AI: Unlimited CEO Bob Elliott
CNBC Television · 16K views
Set up your dot in the ChatGPT mobile app
OpenAI · 84.6K views
OpenAI's $70B Run Rate Meets AI's Infrastructure Challenge
Bloomberg Tech · 13K views
Kleiner Perkins: AI’s Next Trillion-Dollar Companies Go Physical
Bloomberg Tech · 5.4K views
The Rundown: Tech talent pipeline dries up, AI companies hunt for more capital
Bloomberg Tech · 2.2K views
Building the Codex community
OpenAI · 15.3K views
Harvard Psychologist Explains How AI Hacks Your Brain
Future of Life Institute · 4.7K views
How Sophos cuts threat response time by 96% with OpenAI Daybreak
OpenAI · 17.7K views
OpenAI’s “Alien Math” Is Freaking People Out
Wes Roth · 376K views
Production Monitoring with Codex: Grafana, Kubernetes, & Security
OpenAI · 26.7K views
More visual answers from GPT-6 with Intelligent UI
OpenAI · 17.4K views
How Oracle Uses ChatGPT Work to Transform Recruitment
OpenAI · 6.9K views
Get a more visual answer from GPT-6 with Intelligent UI
OpenAI · 8.8K views
Play a game against GPT-6 with Intelligent UI
OpenAI · 13.2K views
Mistral Large 4 Failed My Rubik's Cube Test
Matthew Berman · 2.7K views
12 Grok Bot Use Cases That Feel Illegal
Matthew Berman · 41.6K views
How to Find Nano Banana 2.1 in Google AI Studio
Matthew Berman · 4.8K views
DeepMind's New AI Just Cracked The Code Of Life
Two Minute Papers · 135K views
OpenAI's secret model just BROKE math...
Wes Roth · 141K views
Papers
RSI-Anything: Unlocking Recursive Self-Improvement with World Models
arXiv:2610.13912 · Cui, T. et al.
MC-EvoBench: Auditing Experience-Driven Self-Improvement in Minecraft Agents
arXiv:2610.13829 · Song, J. et al.
EndoRSI: Physical-Evidence-Driven Recursive Improvement of Endoscopic Robot Policies
arXiv:2610.13376 · Ng, C. et al.
VETR: Execution-Grounded Self-Improving Agent for Symbolic Graphics Reasoning
arXiv:2610.13098 · Ma, Y. et al.
Learning More by Training Less: Uncertainty-Guided Progressive Contraction for Efficient Reasoning Post-Training
arXiv:2610.12933 · Yang, S. et al.
Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement
arXiv:2610.12369 · Hu, K. et al.
TestPrism: Rethinking Test Evaluation Beyond a Single Reference
arXiv:2610.12289 · Li, H. et al.
Language Models as AI Research World Models
arXiv:2610.12235 · Wang, Z. et al.
ReSI: Recursive Safety Improvement toward Resistant and Resilient AI
arXiv:2610.12233 · Zheng, J. et al.
DataSense-Bench: The First Step Toward an AI Scientist
arXiv:2610.12190 · Zhang, Y. et al.
Recursive Self-Improvement through Multi-Agent Self-Supervision
arXiv:2610.12176 · Lee, H. et al.
Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks
arXiv:2610.11794 · Zhao, H. et al.
SWE-Journey: Towards More Realistic Evaluation of Coding Assistants through Long-Horizon, Multi-Turn Interaction
arXiv:2610.11559 · Deng, H. et al.
OpenProblemBench: Benchmarking AI on Open Problems in the Foundational Theoretical Sciences
arXiv:2610.11118 · Li, Z. et al.
TraceICL: Thread-Resolved Demonstration Curation with Consistency Calibration for Robust Multimodal In-Context Learning
arXiv:2610.12516 · Yilmaz, Y. et al.
RSIGym: A Flexible Environment for Recursive Self-Improvement
arXiv:2610.10310 · Meng, F. et al.
Lab notes
Impactful scheduling for GPU clusters
Hugging Face
Does better work always mean better workers?
Google Research
Multimodal open d1 decision models for the edge
Hugging Face
Introducing Falcon ASR
Hugging Face
What AI gets wrong and what failure teaches us
Microsoft Research