Apple and a Hacker’s Future

I was happy for years in Apple's walled garden; with AI, however, their protections feel like limitations.

The Billion Dollar AI Advantage Is Disappearing

n8n@2.41.7

OpenAI’s Head of ChatGPT: We’re entering a new era of AI (again) | Tibo Sottiaux

Listen now (37 mins) | Tibo Sottiaux on why Dots is OpenAI’s biggest bet, why agents will dominate internet traffic, and what builders are still getting wrong

v2.1.289

The Agent Said It Was Done. The Database Disagreed.

🧠 Community Wisdom: Getting value from personal AI agents without handing over your inbox, the ego threat of going from IC to manager, uncommon perks to negotiate for, and more

Community Wisdom 301

Legal Center - legal.mistral.ai

<a href="https://news.google.com/rss/articles/CBMiaEFVX3lxTE0td2ZWVVZ4cDRWWVc2RG1zZXBEYndrRHJTVjdzOFFBbHVWRnotVGk2TGppSEsxYkVKY3NKcGh3Y0g4RklqNmdvMmxmSk5lQk1ickQzbzF6bW1VUGhoM1ptMzRuZmREbjNF?oc=5" target="_blank">Legal Center</a>&nbsp;&nbsp;<font color="#6f6f6f">legal.mistral.ai</font>

Claude Frontier Academy: $100M to train 10,000 engineers - Anthropic

<a href="https://news.google.com/rss/articles/CBMiZEFVX3lxTE9FNlFxVHBSRkl5bVF0c253UE5sUEFac3dDZmQ4bzFjc2lER0hNcjFmUUxPVFI5T0xQcldFTXhsd1diR2h4UUVoRnl4TWxEVnp5WXpCMFdkZHNKdlA1bFV4QTVqUDI?oc=5" target="_blank">Claude Frontier Academy: $100M to train 10,000 engineers</a>&nbsp;&nbsp;<font color="#6f6f6f">Anthropic</font>

v2.1.288

2026.40: Dots and Question Marks

The best Stratechery content from the week of September 28, 2026, including Meta's focus, what OpenAI is doing, and Mao and NBA Media Day.

langchain-text-splitters==1.1.3

Open-sourcing AstaBrief, the fast report-generation model in Asta

The Case for the American Manufacturing Asset Class

Rebuilding the industrial base required for deterrence is one of the defining investment opportunities of the next decade.

NVIDIA DGX Spark 64GB Gives Developers More Ways to Build and Scale Local AI

Local AI is becoming more useful by the token. As AI agents move from experiments into everyday development, increasingly capable open models are shrinking to fit on more devices, giving builders more to run locally. Coming this month, NVIDIA DGX Spark will be available with 64GB of unified memory from top manufacturer partners — Acer, […]

n8n@2.41.6

AutoSynthData: Generating Training Data for Enterprise Agents

How NVIDIA GPUs Help Accelerate OpenAI’s GPT-6 Astra Ultrafast

GPT-6 Astra Ultrafast, running on NVIDIA Blackwell GPUs, is available now in the OpenAI API and to eligible ChatGPT Work and Codex users. Accelerated by inference optimizations through OpenAI’s models that tap into the capabilities of the NVIDIA Blackwell architecture, Ultrafast offers up to 8x faster token generation than the Astra Standard mode. For developers, […]

Mistral Studio Specialization issued by Mistral AI to Luis O Kelly - learn.mistral.ai

<a href="https://news.google.com/rss/articles/CBMiZEFVX3lxTFB2eTZMTjh3RXZSSWV0M2lUNkdTbHppNHJJdTNXd2RYd0dwZnRDT1I4RU5vUjEzcURLVTAwdXFtZjVPSVViWHh2R3F4Z2cxSWo0ZS1OR1RnNkZmZTZzYWVwQmt6aG8?oc=5" target="_blank">Mistral Studio Specialization issued by Mistral AI to Luis O Kelly</a>&nbsp;&nbsp;<font color="#6f6f6f">learn.mistral.ai</font>

v2.1.287

Vivian Midha Shen and Raphael Schaad Join YC as General Partners

We’re thrilled to announce that Vivian Midha Shen and Raphael Schaad are joining Y Combinator as our newest General Partners.

A big-tent or small-tent AI safety movement?

The unstated disagreement that underpins safety debates

Investing in doxxnet

a16z leads doxxnet’s Series A

Barclays scales Claude to upgrade operations and improve client experience - Anthropic

<a href="https://news.google.com/rss/articles/CBMiY0FVX3lxTE9iWlNWRnpuZHlvMWUyZ3ZXanVqU1NjQ3FaaTlpVHlMaDhzdzdub25CMERKYThTYjN5RFFaSzBlZGJEbHNNMkpLN0M1Wkhfa2ltdFJZX0dpY3BJNndZLXhuRHN1Zw?oc=5" target="_blank">Barclays scales Claude to upgrade operations and improve client experience</a>&nbsp;&nbsp;<font color="#6f6f6f">Anthropic</font>

Claude-shaped science - Anthropic

<a href="https://news.google.com/rss/articles/CBMiZ0FVX3lxTE1UWno0NFZ1Y0JIWFRYSFZoRE9iVmx5Z093dlk0aXVQc2JHSVFocFZuelQ3azV3Mm1vcnhiLXZ5d213R0RPc29ia0NYd1VvTUJjWWZUWi1qVHItQmRnVFRyZzRUWENXV28?oc=5" target="_blank">Claude-shaped science</a>&nbsp;&nbsp;<font color="#6f6f6f">Anthropic</font>

Investing in Armadin

a16z leads Armadin’s Series B

An Interview with Jason Del Rey About Muse, Amazon, and Walmart

An interview with Jason Del Rey about Amazon versus Meta, which is a continuation of the oldest battle in retail between Amazon and Walmart.

n8n@2.42.2

n8n@2.41.5

How the bad science of AI doomerism is good for big business

AI Now's Senior Research Scientist, Boyan Milanov, points out how AI doomerism isn't based on any scientific work and that calls for regulation from within the industry are, in reality, attempts by AI companies to set their own standards. The post How the bad science of AI doomerism is good for big business appeared first on AI Now Institute .

White House Releases Taxpayer-Funded Midterm Ads

In the wake of NVIDIA's announcement of a new agent safety platform, AI Now's Aya Ibrahim joins "Balance of Power" to warn against "one-size-fits-all" security steps. The post White House Releases Taxpayer-Funded Midterm Ads appeared first on AI Now Institute .

AI Almost Started a U.S.–China War — and No One Seems to Care

Despite concerns and warnings about the harms of their technology, AI companies are still pushing for and promoting their products as instrumental in national infrastructure. AI Now's Heidy Khlaaf says that acknowledging these issues wouldn't be profitable for AI companies. The post AI Almost Started a U.S.–China War — and No One Seems to Care appeared first on AI Now Institute .

AI existential risk probabilities are (still) too unreliable to inform policy

How speculation gets laundered through pseudo-quantification

Seed 模型家族 - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMiX0FVX3lxTE5EamxmSkpTQll1NW0wRFdKX0hUTXdwclU2VEY5MGVGZ25rZ0pMUTRmdXJwTTVRaDB0RjVQTklINFZBdUVyNENzaGpGY1NKNVlfQ2wtdk94T0xVS3hzLWRJ?oc=5" target="_blank">Seed 模型家族</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteDance Seed</font>

The Phone Number puzzle

The last IMO problem AI could not solve

The last IMO problem AI could not solve

The AI-as-Normal-Technology view of loss-of-control incidents

A middle ground between the cybersecurity and AI safety communities

Audio Prompting - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMiZ0FVX3lxTE5lbUV3X1VDT2JLZGJWcjByRjZJNFUwM2ktOVdrVGN6VmhPbXNsR1JyRUJ5SEtLeVVvWC1PYm05bU1WYlZQSHRDMUdOT0VVWEExNElWT3NOenA4RXZqRzg1N0h0a1REVWs?oc=5" target="_blank">Audio Prompting</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteDance Seed</font>

研究进展- 字节跳动Seed - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMiVkFVX3lxTE9SbWR2ZXJoMzlpWElLUjZpZGs5NWhyNUdtTFVfQmxxYlBfRU5fcTgzejA0NHc0Q3ZQLUM3OWhqOWM0OFNrWlZ5S3ZQV1hsemp2cDVNNEln?oc=5" target="_blank">研究进展- 字节跳动Seed</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteDance Seed</font>

博客 语音智能 - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMiWEFVX3lxTE1vTVhYTDFuLVdvNkhDQzNqZFAwckdZVXFibWUzeFdpeE9tcWZ4akFQcWxfV09ERnRwdnNiZmdEMlV5UEdOLUlrQmEzcGdpcVVRTXZyUm1GbWg?oc=5" target="_blank">博客 语音智能</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteDance Seed</font>

团队动态- 字节跳动Seed - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMinwFBVV95cUxPT0lBQWFzZ1hjZlN4eGNRSzR5ZVR3RGdpOEVETnRQRVozZ0MyMUlZTWlURDVSMF9sQWN3eDlETmxEZVFJRlVtUEsxQlFNOHd5MlYtaXBudWR1aFk4Qzl2S1FQR1gyUDRhSTlSYmdGbUlmaE5uWEllOFBvbnRmZHZfNFBhV3NXM2JCOU1QOE54bEl5N2VpNi1KeVU2VHp4QWc?oc=5" target="_blank">团队动态- 字节跳动Seed</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteDance Seed</font>

从“会说”走向“会创作”｜ Seed Audio 1.0 音频创作模型发布 - ByteDance Seed

<a href="https://news.google.com/rss/articles/CBMitwFBVV95cUxOVVg3YlgxWUljaHZHb3c5LUlhVlRtekg2MzBBcWpWd2xHRTE4VllVQnd1VUx0emd2ZjVGcWNKZHBTdmtjY2RVaWZZVWVoWDNvdEcxb3B0NG9BdXp3Wmo1N0VTTXltbWVZZ2t0QkNVY2JNczhxeXRMUmlXSW5tOWFfSGFiOGNnZ0FDWklfblNJSEtOSkhmRG9NMlBVbldIWjgyX1l2elFsMmpNRG9zLXJZeEIzc0JHeDA?oc=5" target="_blank">从“会说”走向“会创作”｜ Seed Audio 1.0 音频创作模型发布</a>&nbsp;&nbsp;<font color="#6f6f6f">ByteD

Harness Engineering for Self-Improvement

<p>The concept of <strong>recursive self-improvement (RSI)</strong> dates back to <a href="https://philpapers.org/rec/GOOSCT">I. J. Good (1965)</a>, where he defined an &ldquo;ultraintelligent machine&rdquo; as a system that can surpass humans in all intellectual activities and design better machines to improve itself. <a href="https://www.lesswrong.com/posts/JBadX7rwdcRFzGuju/recursive-self-impro

Scaling Laws, Carefully

<p>Scaling laws are one of the most critical empirical findings in deep learning. The observation is simple in form: the training loss $L$ decreases predictably as we scale up model size $N$, dataset size $D$, and compute $C$, following a power-law curve, which appears as a straight line on a log-log plot. We can view scaling laws as a framework for describing the relationship between compute, los