Turing Post

Then came the contradiction: closed models blocked Hugging Face from analyzing the attack evidence, so its security team turned to the open-weight GLM-5.2!

We explain what happened, who was in control and why this incident may push OpenAI with help from Hugging Face toward AI that is both more open and safer

Attention Span explains the technical and business choices shaping AI.

👉 Subscribe for high-signal AI mechanics 

👉 Into videos? Check our IG / turingpost_tv and TikTok / turingpost_tv 👉 More analysis: TuringPost.com 

👉 Interviews: @realturingpost

*Links:*

Hugging Face: https://huggingface.co/blog/security-incident-july-2026

OpenAI: https://openai.com/index/hugging-face-model-evaluation-security-incident/

ExploitGym: https://arxiv.org/abs/2605.11086

Adrien Carreira: https://x.com/XciD_/status/2079191130248212591

Simon Willison: https://simonwillison.net/2026/Jul/22/openai-cyberattack/

Thomas Ptacek: https://x.com/tqbf/status/2080290141063569791

LeRobot: https://x.com/LeRobotHF/status/2080296144832274881 

#AI #OpenAI #HuggingFace #OpenSourceAI #Cybersecurity

In this episode, we explain why serving an agentic model is so expensive, how a GPU shortage became an IPO pitch, what the July 27 weights release changes, and what a ban of an open model can and cannot actually reach.

Attention Span is here to explain the technical and business choices shaping AI.

👉 Subscribe for high-signal AI mechanics 👉 

Into videos? Check our IG https://www.instagram.com/turingpost_tv and TikTok https://www.tiktok.com/@turingpost_tv 

👉 More analysis: TuringPost.com 

👉 Interviews: @realturingpost

*Links:* 

Kimi K3: The open-weights escalation https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation

Kimi.ai capacity announcement https://x.com/Kimi_Moonshot https://x.com/Kimi_Moonshot/status/2078855608565207130

Xi Jinping's Big AI Speech, Annotated https://mattsheehan.substack.com/p/xi-jinpings-big-ai-speech-annotated

Kimi K3: The open-weights escalation https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation

Reuters, Moonshot pauses subscriptions amid IPO push SCMP, Kimi K3 developer suspends new subscriptions The Decoder, membership split into two tiers Qwen 3.8 announcement https://x.com/Alibaba_Qwen/status/2078759124914098291 MarkTechPost, Qwen 3.8 preview without benchmarks or license

Reuters, Xi's WAIC keynote Quartz, WAICO launch and membership Axios (Maria Curi), the administration's signals on Chinese models Nathan Lambert, Interconnects: Notes from inside China's AI labs https://www.interconnects.ai Previous episode: Kimi K3 and Inkling

#AI #ArtificialIntelligence #OpenSourceAI #KimiK3 #Qwen #LLM #MachineLearning #TechNews #China #AIPolicy

Kimi K3 is a 2.8-trillion-parameter Chinese model that jumped from #18 to #1 in the Frontend Code Arena, ahead of Claude Fable 5. WHAT?! 

Inkling is the first major model from Mira Murati’s Thinking Machines Lab, and the company states directly that it is not the strongest model available. WHAT?!

Both strategies make sense once we examine what the companies are building. 

In this episode, we explain K3’s 896-expert architecture, why Moonshot recommends at least 64 accelerators, how LoRA customization works, what community quantization changed for Inkling, and which organizations gain meaningful control from open weights.

Attention Span is here to explain the technical and business choices shaping AI.

👉 Subscribe for high-signal AI mechanics 

👉 Into videos? Check our IG https://www.instagram.com/turingpost_tv and TikTok https://www.tiktok.com/@turingpost_tv 

👉 More analysis: TuringPost.com 

👉 Interviews: @realturingpost

Links:

About LoRA https://www.turingpost.com/p/lora 

Moonshot AI: The Chinese Unicorn Revolutionizing Long-Context AI https://www.turingpost.com/p/moonshotai 

Thinking Machines Lab, “Inkling: Our open-weights model” https://thinkingmachines.ai/news/introducing-inkling/ 

Thinking Machines Lab, Inkling Model Card https://thinkingmachines.ai/model-card/inkling/ 

Thinking Machines Lab, Tinker https://thinkingmachines.ai/tinker/ 

Moonshot AI, “Kimi K3: Open Frontier Intelligence” https://www.kimi.com/blog/kimi-k3 

Arena, Kimi K3 Frontend Code Arena result https://x.com/arena/status/2077824029126504525  

Semianalysis about Kimi K3 https://x.com/SemiAnalysis_/status/2077966560447074689 

Arena, Code Arena methodology https://arena.ai/blog/code-arena/ 

Artificial Analysis, Kimi K3 https://artificialanalysis.ai/models/kimi-k3 

Artificial Analysis, Inkling https://artificialanalysis.ai/articles/thinking-machines-has-released-inkling-the-new-leading-u-s-open-weights-model 

Unsloth, community Inkling quantizations https://unsloth.ai/docs/models/inkling

#AI #ArtificialIntelligence #OpenSourceAI #LLM #MachineLearning #GenerativeAI #KimiK3 #MiraMurati #AIModels #TechNews

Sarah Bird, Chief Product Officer of Responsible AI at Microsoft, has been in this field since it was a niche. Now it's everywhere. In this episode of Inference, she explains why the entire software development lifecycle is changing – and why human oversight has to evolve with it.

*In this episode of Inference, we get into:*

- Why "responsible AI" is the wrong framing – and why "trustworthy AI" matters more 

- What changes when agents write the code AND review the code 

- Why open-sourcing responsible AI tools is non-negotiable – "we don't want to be competing on this" 

- The dual-use problem: AI that finds vulnerabilities helps defenders AND attackers 

- Why fixing it "in the model" sounds easy but doesn't work – and who's actually responsible at each layer 

- The shift from chat to agentic to physical AI – and why consequences scale dramatically 

- What kids should learn about AI: the "stoplight" framework from New South Wales schools 

- Why Sarah is surprisingly optimistic – and why low p(doom) is a rational position 

- The three problems her team is solving right now: emerging risks, agent governance, and the new software lifecycle

We also talk about regulation, why responsible AI requires linguists alongside engineers, the rise of "psychosocial risk," and why humans + AI is more exciting than AGI alone.

This is a conversation about what it actually takes to make AI systems we can trust – and why humans are not optional. *Watch it!*

*Guest:* 

Sarah Bird, Chief Product Officer of Responsible AI at Microsoft https://www.linkedin.com/in/slbird/ 

https://x.com/slbird   

https://www.microsoft.com/en-us/ai/responsible-ai 

*Did you like the episode?*

📌 Subscribe for more conversations with the builders shaping real-world AI 

💬 Leave a comment if this resonated 

👍 Like it if you liked it 

🫶 Thank you for watching and sharing!

📰 Want the transcript and edited version? Subscribe to Turing Post: https://www.turingpost.com/subscribe

*Chapters:*

0:00 Is Responsible AI Possible?

1:21 The Pace of AI Innovation

3:10 Tools: Assert & Agent Control

5:21 Defining Trustworthy Contexts

6:38 Diverse Teams & Cross-Domain Work

8:16 The Need for Regulation

11:37 Defining AGI & Human-in-the-Loop

13:55 Limits of Agent Delegation

16:12 Generative vs. Physical AI

17:25 Addressing Accountability

20:36 User Responsibility & Best Practices

22:01 AI Literacy for Children

23:28 Democratizing Innovation

29:21 Three Strategic Focus Areas

31:37 Book Recommendation: The Culture Map

*Turing Post* is a newsletter about AI's past, present, and future. Publisher Ksenia Se explores how intelligent systems are built – and how they're changing how we think, work, and live.

Sign up: https://www.turingpost.com

Follow us - https://x.com/TheTuringPost https://www.linkedin.com/in/ksenia-se https://huggingface.co/Kseniase

#ResponsibleAI #Microsoft #AIAgents #AGI #TrustworthyAI #AIGovernance #AIRisk #SarahBird

What happened to GitHub then? Record acceleration across commits, PRs, Actions, and security scans – and a fundamental rethink of what GitHub even is.

*In this episode of Inference, we get into:*

The December 2025 capability jump

GitHub's massive scale challenge

"Low floor, high ceiling" – why lowering the barrier to creation may be the biggest GDP unlock in history

The Mozart problem: how many geniuses never had access to a piano – and why AI changes that

From UI → UX → AX:

The new Compiler app and "canvases" – bidirectional surfaces where humans and agents co-create in real time

Why Mario is anti-parallelization hype: "You could parallelize yourself to no value at all"

Macro vs. micro delegation

Why CoPilot will always be co-pilot, not pilot – and where the human stays in the loop

We also talk about the redefinition of "developer," why creation (not efficiency) drives human progress, and how GitHub plans to serve both the first-time builder and the Picasso-level craftsman on the same continuum.

This is a conversation about the future of software, the role of the human, and what it means when everyone becomes a builder. Watch it!

*Guest:*

Mario Rodriguez, Chief Product Officer at GitHub:

https://www.linkedin.com/in/mariorodriguez3/

https://x.com/mariorod1

https://github.com/mariorod

*Chapters:*

0:00 Intro — GitHub's Agentic Future

0:35 What Changed When AI Agents Started Working

4:41 The Engineering Challenges of Explosive AI-Driven Growth

6:59 GitHub's New Mission: Lower the Floor, Raise the Ceiling

9:58 Why AI Will Create More Builders, Not Fewer Developers

11:44 From UI to AX: Designing an Agent-Native GitHub

14:54 Is Everyone a Developer Now?

17:54 Advice for Young Developers in the AI Era

19:28 The Art of Micro-Delegation with AI Agents

21:14 Copilot Pricing, Token Costs, and Smarter AI Usage

24:40 AGI, Human Progress, and the Future of Creation

27:22 Why Humans Will Stay in the Loop Forever

29:24 The Books and Ideas That Shaped Mario Rodriguez

*Follow on*: https://www.turingpost.com/

📖 Related reading on turingpost.com:

→ AI Agents Vocabulary: https://www.turingpost.com/p/agentsvocabulary

→ The AI Software Stack Explained: https://www.turingpost.com/p/aisoftwarestack

→ Inside Cognition — Devin and the Coding Agent Era: https://www.turingpost.com/p/cognition

*Did you like the episode? You know the drill:*

📌 Subscribe here and here (https://www.turingpost.com/subscribe) for more conversations with the builders shaping real-world AI.

💬 Leave a comment

👍 Like it

🫶 Thank you for watching and sharing!

#GitHub #AI #AIAgents #Developers #Copilot #VibeCoding #AgenticAI #futureofcoding #githubcopilot #AIagents2026 #GitHubCPO #agenticcoding #futureofdevelopers #GitHubAI #codingagents #GitHub2026 #turingpost

Ammaar Reshi leads Product and Design at Google AI Studio (DeepMind). His path is unusual: from writing iPhone app reviews as a teenager, to Palantir, Brex, and ElevenLabs, to now designing how millions of people build apps with AI in Google. His philosophy is simple – nothing is impossible, it's just time and iteration, and you learn by making and showing your work openly. Ammaar has a great energy and insights into how to stay relevant with AI.

*In this episode of Inference, we get into:*

How "tap, tap, tap" prompting helps anyone write rich specs without knowing what an NPM package is

Why the chat interface is evolving into something more like Slack – with daily standups with your agents

The case for abstraction: hiding Firebase, OAuth, and the messy parts from people who just want to build

Why speed (not intelligence) is now the real bottleneck for AI building

The gap between a stunning demo and a product that serves billions

How Ammaar is retraining Google itself – PMs across the company learning to vibe code and come up with amazing ideas

Why Ammaar open-sources almost every demo he builds – and how that inspires the next wave of builders

Ammaar's definition of AGI and lessons from Steve Jobs and Aurelius

A conversation about building, designing, and the new relationship between humans and intelligent tools. Watch it!

Did you like the episode? You know the drill: 📌 Subscribe for more conversations with the builders shaping real-world AI 💬 Leave a comment if this resonated 👍 Like it if you liked it 🫶 Thank you for watching and sharing!

Guest: Ammaar Reshi, Lead Product + Design at Google AI Studio (DeepMind) https://www.linkedin.com/in/ammaarsreshi https://x.com/ammaar https://ammaar.me

📰 Want the transcript and edited version? Subscribe to Turing Post: https://www.turingpost.com/subscribe

Turing Post is a newsletter about AI's past, present, and future. Publisher Ksenia Se explores how intelligent systems are built – and how they're changing how we think, work, and live.

Sign up: https://www.turingpost.com

Follow us - https://x.com/TheTuringPost https://www.linkedin.com/in/ksenia-se https://huggingface.co/Kseniase

#AI #GoogleAIStudio #DeepMind #VibeCoding #AIAgents #AGI #Gemini

Eric Ries, entrepreneur and author of The Lean Startup and Incorruptible, argues that the more valuable a company becomes, the more pressure it faces to make money without creating real value. In other words: corruption in business doesn’t begin at the margins. It begins when builders stop protecting what made their company worth building in the first place.

*In this episode of Inference, we get into:*

- Why every successful company eventually becomes a target

- What “corruption” really means in business

- Why it’s always too early to protect your mission – until it’s too late

- How trust and love become real competitive advantages, not just soft language

- What Eric means by “financial gravity” and how companies get pulled into mediocrity

- How mission-driven companies can “bend the light” and shape the economy around them

- Why younger generations no longer trust institutions – and what builders must replace

- How AI will accelerate institutional collapse if we use it only for cost-cutting

- Whether AI could become a “mission guardian” inside companies

- What it would mean to redefine capitalism around human flourishing

We also talk about private equity, broken incentives, collapsing institutions, the civic consequences of replacing old systems with new technologies, and why the next generation of founders may have to rebuild far more than startups – including the trust infrastructure of society itself.

This is a conversation about business, power, values, and what it takes to build something that can survive success without losing its soul. It’s truly a must watch.

*Guest:*

Eric Ries, entrepreneur, author of The Lean Startup and Incorruptible

🔗 LinkedIn: https://www.linkedin.com/in/eries

🔗 X / Twitter: https://x.com/ericries

🔗 Incorruptible (new book): https://incorruptible.co

🔗 Lean Startup: https://leanstartup.co

#EricRies #Incorruptible #LeanStartup #AI #AGI #Startups #Entrepreneurship #Capitalism #Trust #Innovation

*Turing Post* is a newsletter about AI’s past, present, and future. Publisher Ksenia Se explores how intelligent systems are built – and how they’re changing how we think, work, and live.

📖 Why AI transformation is harder than it looks: https://www.turingpost.com/p/orgage1

Follow us

Ksenia and Turing Post:

https://x.com/TheTuringPost

https://www.linkedin.com/in/ksenia-se

https://huggingface.co/Kseniase

*Did you like the episode? You know the drill:*

📌 Subscribe here and here (https://www.turingpost.com/subscribe) for more conversations with the builders shaping real-world AI.

💬 Leave a comment

👍 Like it and – I don’t know how it works – hype it?

🫶 Thank you for watching, sharing and leaving your comments! Love talking to you.

Clément Delangue, co-founder and CEO of Hugging Face, believes we are entering a new phase of AI – one where building models, fine-tuning systems, running local AI, and even experimenting with robotics may no longer be limited to a small technical elite.

His passion for open source is very contagious! I enjoyed chatting with him about:

Why the next wave of AI builders won't be traditional engineers – and how that could push the field beyond slop toward biology, medicine, and climate

What open source actually solves in cybersecurity – and why "safety" is often a cover story for business strategy

Why lobbying against open source in the US would be a strategic mistake that could cost the country its AI leadership

Why comparing open weights to closed APIs is irrelevant and why benchmarks miss what really matters

What Hugging Face is learning as agents become a new kind of user

How LeRobot and Reachy Mini are turning AI into something people lile

Why training, fine-tuning, and post-training on your own data are becoming the real differentiators as building apps gets trivial

What three months of paternity leave taught Clem

We also talk about fear-based AI marketing, how public perception shifts the moment people build with AI, what's missing in robotics datasets, and why Clem keeps coming back to Camus' Sisyphus as a metaphor for being a founder right now

A conversation about agency, openness, and what it means to democratize AI before it gets locked down. Watch it.

*Chapters:*

00:00 AI Builders Are About to Explode

00:35 Why Coding Agents Still Struggle with AI

02:23 100 Million AI Builders

03:30 Non-Technical People Entering AI

05:15 How Building AI Can Change Public Perception

06:22 Who Can Make AI More Open?

08:02 Fear-Based Marketing in AI

09:33 Open Source, Cybersecurity, and Risk

12:31 Why Companies Don’t Open Source

14:30 Lobbying Against Open Source

17:24 What Changed During Paternity Leave

19:11 Making Hugging Face Agent-Native

21:00 Hugging Face Robotics and LeRobot

23:01 Local AI, Open Models, and the Future

*Did you like the episode? Do the following:*

📌 Subscribe here and here (https://www.turingpost.com/subscribe) for more conversations with the builders shaping real-world AI.

💬 Leave a comment

👍 Like it

🫶 Thank you for watching and sharing!

*Guest:*

Clément Delangue, co-founder & CEO of Hugging Face

https://x.com/ClementDelangue

https://www.linkedin.com/in/clementdelangue

https://huggingface.co/clem

https://huggingface.co/

*Projects discussed:*

ML Intern

LeRobot

SO-101 / LeRobot docs

📰 Want the transcript and edited version?

Subscribe to Turing Post: https://www.turingpost.com/subscribe

Turing Post is a newsletter about AI’s past, present, and future. Publisher Ksenia Se explores how intelligent systems are built – and how they’re changing how we think, work, and live.

Follow us - Ksenia and Turing Post:

https://x.com/TheTuringPost

https://www.linkedin.com/in/ksenia-se

https://huggingface.co/Kseniase

#HuggingFace #ClemDelangue #OpenSourceAI #LocalAI #AIAgents #RoboticsAI #LeRobot #MLIntern #AIBuilders #FutureOfAI

Neeru Khosla, co-founder of CK-12 Foundation, believes this moment could become a turning point for education. After nearly two decades building free learning tools for students and teachers, she argues that AI is our chance to finally understand how students think, where they get stuck, and how to help each child learn in a way that works for them.

*In this episode of Inference, we get into:*

- Why prompting is not cheating, but a real learning skill

- Why textbooks alone were never enough for deep understanding

- What C means in CK-12 (it’s important!)

- What Neeru learned from launching Flexi, CK-12’s AI tutor, now used by millions of students

- Why standardized testing misses the most important part of learning

- Why teachers need support, visibility, and confidence – not fear

- Why AI literacy may become as fundamental as reading, writing, and math

- How curiosity, mentorship, and community shape better learning outcomes

- Why “attention is all you need” is no longer enough – and what we need now

We also talk about public school inertia, philanthropy, core values when raising kids, and why Neeru believes AI should be used as augmented intelligence – not something to fear, but something to help humans grow.

This is a conversation about education, equity, curiosity, and what it would really take to build a learning system that works for every child. Watch it. I’m really passionate about this topic and think that everyone should think and talk more about it.

*Did you like the episode? You know the drill:*

📌 Subscribe for more conversations with the people rethinking how AI will shape society

💬 Leave a comment if this resonated with you

👍 Like it if you liked it

🫶 Thank you for watching and sharing

*Guest:*

Neeru Khosla, co-founder and executive director of CK-12 Foundation

https://info.ck12.org/neeru-khosla

https://www.ck12.org/flexi/

*📰 Want the transcript and edited version?*

Subscribe to Turing Post: https://www.turingpost.com/subscribe

*Chapters*

0:00 Why Education Is the Greatest Gift to Society

0:27 Meet Neeru Khosla & the Mission of CK-12

1:32 How Technology Changed Learning Over the Years

3:20 Why AI Became a Turning Point in Education

5:52 Flexi: CK-12’s AI Tutor Used by Millions

8:26 Does AI Require Rethinking the Education System?

11:12 What Teachers Need From AI Right Now

12:56 Essential Skills for Kids and Teachers in the AI Era

18:13 From Molecular Biology to Building CK-12

23:27 Why Education Is a Human Right — and What Comes Next

Turing Post is a newsletter about AI’s past, present, and future. Publisher Ksenia Se explores how intelligent systems are built – and how they’re changing how we think, work, and live.

*Follow us*

Ksenia and Turing Post:

https://x.com/TheTuringPost

https://www.linkedin.com/in/ksenia-se

https://huggingface.co/Kseniase

#AIinEducation #EducationAI #NeeruKhosla #CK12 #Flexi #AILiteracy #FutureOfEducation #EdTech #PersonalizedLearning #TuringPost

If you want to understand where AI goes next and what occupies the minds of the best researchers, you need to watch this video.

*In this episode:*

Why transformers and world models are not competing ideas

Why physical AI is still a major frontier

The evolution of simulation

Why 3D matters for robotics and real-world intelligence

What’s still missing in multimodal AI

Whether autonomous driving could have a “ChatGPT moment” before robotics does

If you enjoy conversations at the edge of AI research, *subscribe to Turing Post* for more interviews with the people building the future https://www.turingpost.com/

*Chapters:*

0:00 Physical AI vs Transformers — The Big Question

0:19 Introduction: NVIDIA & Spatial Intelligence Lab

0:38 Transformers vs World Models — Not a Competition

1:45 World Models as Simulators of Reality

3:20 Are New Architectures Replacing Transformers?

4:17 “Alpa Dreams” — Real-Time Interactive AI Worlds

6:22 The Evolution of Simulation in Self-Driving

7:44 From 3D Reconstruction to True World Modeling

10:26 Multimodal AI: Audio, Radar, and Physical Interaction

13:29 AGI, Robotics & the Future of Physical AI

*Did you like the episode? You know the drill:*

📌 Subscribe here and here (https://www.turingpost.com/subscribe) for more conversations with the builders shaping real-world AI.

💬 Leave a comment

👍 Like it

🫶 Thank you for watching and sharing!

*Guest:*

Sanja Fidler – NVIDIA Research https://research.nvidia.com/person/sanja-fidler

University of Toronto https://www.cs.toronto.edu/~fidler/

Spatial Intelligence Lab https://research.nvidia.com/labs/sil/

Google Scholar https://scholar.google.com/citations?user=CUlqK5EAAAAJ&hl=en

X https://x.com/FidlerSanja

LinkedIn https://ca.linkedin.com/in/sanja-fidler-2846a1a

#AI #NVIDIA #SanjaFidler #WorldModels #PhysicalAI #SpatialIntelligence #Robotics #AutonomousDriving #Transformers #GTC