Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Links / Past days

Friday, 18 September 2026

The AI links the most newsletters and link sources had shared by the end of the day (UTC), each source counted once.

As of 2026-09-18 23:59 UTC Status sealed Rebuilt from issues on file

Reconstructed after the day from the issues on file (dates as published). An issue counts on the date it carries, even if it was collected later.

That day 36-hour window

  1. Shared by 4

    OpenAI’s new rules for reporting model misbehavior

    “OPENAI📂 OpenAI’s new rules for reporting model misbehaviorImage source: OpenAIThe Rundown: OpenAI just released six new reports on its models misbehaving during training, including wild details like rewriting their own jailbreak-style instructions to using leaked credentials, alongside a new process for making such incidents public faster.The details: An…” via The Rundown AI

    Reported byDon't Worry About the Vase, Ben's Bites, The Rundown AI, Superhuman AI

    How 1 source put it
    • Don't Worry About the Vase2026-09-17issue ↗
    • Ben's Bites2026-09-17issue ↗
    • The Rundown AI2026-09-18issue ↗ OPENAI📂 OpenAI’s new rules for reporting model misbehaviorImage source: OpenAIThe Rundown: OpenAI just released six new reports on its models misbehaving during training, including wild details like rewriting their own jailbreak-style instructions to using leaked credentials, alongside a new process for making such incidents public faster.The details: An…
    • Superhuman AI2026-09-18issue ↗
    openai.com First shared 2026-09-17
  2. Shared by 5

    Introducing System One Models & Jev

    TypeSafe AI launches Jev, a System One model that makes structured decisions 40x to 200x faster than existing LLMs. Includes a pricing table showing $0.042 input costs and free outputs, plus workflow evals comparing Jev against GPT-5.6 Terra. cited

    Reported byThe Rundown AI, TLDR AI, AI Breakfast, Ben's Bites, Superhuman AI

    Where it was shared
    typesafe.ai announcement · 10 min read First shared 2026-09-16
  3. Shared by 3

    Toward Recursive Self-Improvement: How GLM Built Its Own Inference Infrastructure

    “Z AI revealed that its GLM-5.3 AI helped set up the 100K-chip system now serving GLM-5.3-Flash, writing that “our successors are the AI systems we are creating ourselves.”” via The Rundown AI

    Reported byThe Rundown AI, TLDR AI

    Discussed onHacker News 411 pts · 285 comments

    How 1 source put it
    • The Rundown AI2026-09-18issue ↗ Z AI revealed that its GLM-5.3 AI helped set up the 100K-chip system now serving GLM-5.3-Flash, writing that “our successors are the AI systems we are creating ourselves.”
    • TLDR AI2026-09-18issue ↗
    z.ai First shared 2026-09-17
  4. Shared by 2

    Projects redesigned: from folder to conversation

    “🗂️ Projects - Workspace where Claude splits a goal into side-by-side sessions” via The Rundown AI

    Reported byThe Rundown AI, TLDR AI

    How 1 source put it
    claude.com First shared 2026-09-18
  5. Shared by 2

    Astra for Law

    “Astra for Law is a legal AI foundation that combines GPT-6 Astra with tools, settings, privacy controls, and context designed for professional legal work.” via TLDR AI

    Reported bySuperhuman AI, TLDR AI

    How 1 source put it
    • Superhuman AI2026-09-18issue ↗
    • TLDR AI2026-09-18issue ↗ Astra for Law is a legal AI foundation that combines GPT-6 Astra with tools, settings, privacy controls, and context designed for professional legal work.
    openai.com First shared 2026-09-18
  6. Shared by 5

    Claude Cowork and chat are now one Claude

    “Anthropic combined Cowork and Chat into one main Claude app, while also launching its built-in Word and PowerPoint versions called Claude Docs and Slides in beta.” via The Rundown AI

    Reported byThe Rundown AI, Superhuman AI, TLDR AI, Ben's Bites

    Discussed onHacker News 234 pts · 225 comments

    How 1 source put it
    claude.com 3 min read First shared 2026-09-16
  7. Shared by 2

    Qwen3.8-Omni-Flash: Omni Senses. Agentic Delivery

    “Qwen3.8-Omni-Flash is a native omnimodal model. It supports a 1M-token context window with text, image, audio, and video inputs. Qwen3.8-Omni-Flash achieves audio-visual performance close to Gemini 3.8 Flash and overall audio performance that exceeds Gemini 3.8 Flash. It is now available on…” via TLDR AI

    Reported byTLDR AI

    Discussed onHacker News 346 pts · 138 comments

    AboutQwen3.8-Flash-Next, Qwen3.8-2.4T-A95B

    qwen.ai First shared 2026-09-17
  8. Shared by 2

    LLM Classification Is Feature Engineering

    “Getting LLMs into shape to reliably serve as classifiers is hard work but potentially highly impactful. More and more research is relying on LLMs for classification, so these tools need to output high-quality results. This post shows how classification is just a feature engineering problem.” via TLDR AI

    Reported byTLDR AI

    Discussed onHacker News 116 pts · 25 comments

    minimallysufficient.com First shared 2026-09-17
  9. Shared by 1

    How Claude is uplifting biomolecular modeling

    “Claude enhanced biomolecular modeling by optimizing over 30 models, achieving a 4x speed increase and creating a low-memory mode for predicting larger systems on a single NVIDIA GPU. These improvements, now open-sourced, allow for efficient protein design and structure prediction, potentially…” via TLDR AI

    Reported byTLDR AI

    anthropic.com First shared 2026-09-18
  10. Shared by 1

    Google wants to give your family its own cloud computer

    “Google's family agent runs on its own cloud computer and Google account, turning the emails, files, and calendars your household shares into daily briefings and updated plans. It can fill out forms and coordinate activities for up to six people, asking permission before acting outside…” via TLDR AI

    Reported byTLDR AI

    blog.google First shared 2026-09-18
  11. Shared by 8

    Countering misuse of AI: September 2026 / Anthropic

    “Anthropic's Threat Intelligence team has identified and disrupted several operations in the past several months where threat actors tried to use Claude for malicious activity. This report shares case studies from those operations and describes how malicious use of Claude has evolved since…” via TLDR AI

    Reported byThe Rundown AI, Superhuman AI, TLDR AI, AI Breakfast, Interconnects, Ben's Bites, Last Week in AI, The Batch

    How 1 source put it
    anthropic.com 156 min read First shared 2026-09-11
  12. Shared by 3

    What’s new: Meta introduced Muse, a personal AI agent based on the Muse Spark 1.3 model.

    “Meta unveiled its first personal agent: Muse connects to your personal accounts to carry out tasks on your behalf, but some users have raised privacy concerns.” via Superhuman AI

    Reported bySuperhuman AI, TheSequence, The Batch

    How 1 source put it
    about.fb.com First shared 2026-09-11
  13. Shared by 4

    On the Navier–Stokes Millennium Prize Problem

    “Quartz: OpenAI cracks the Navier–Stokes Millennium Prize problemOpenAI fought dirty on career-making math problem, says NYU mathematicianOpenAI’s feud with mathematicians is only escalating” via Last Week in AI

    Reported bySuperhuman AI, TheSequence, Last Week in AI, The Batch

    How 1 source put it
    openai.com First shared 2026-09-11
  14. Shared by 2

    Buckmaster said he and Alpöge had spent most of a year using OpenAI Codex and Anthropic Claude to chase the same kind…

    “Beyond this, if Professor Buckmaster’s claims are true that OpenAI mobilized an internal team and model on the same narrow problem, after a year of his and others’ work inside Codex3, without clear disclosure about overlap or data use, we have to wonder how innovation and discovery can continue while trust and openness degrade.” via Exponential View

    Reported byExponential View, The Batch

    How 1 source put it
    • Exponential View2026-09-13issue ↗ Beyond this, if Professor Buckmaster’s claims are true that OpenAI mobilized an internal team and model on the same narrow problem, after a year of his and others’ work inside Codex3, without clear disclosure about overlap or data use, we have to wonder how innovation and discovery can continue while trust and openness degrade.
    • The Batch2026-09-18issue ↗
    cims.nyu.edu First shared 2026-09-13
  15. Shared by 2

    OpenAI fought dirty on career-making math problem, says NYU mathematician

    “OpenAI fought dirty on career-making math problem, says NYU mathematician” via Last Week in AI

    Reported byLast Week in AI, The Batch

    How 1 source put it
    techcrunch.com First shared 2026-09-17
  16. Shared by 1

    Download the app on iOS and Android

    “Your desk shouldn’t be the only place where you can learn. The DeepLearning.AI app lets you follow the latest news and track your course progress anywhere and everywhere, on your commute or coffee break. Download the app on iOS and Android” via The Batch

    Reported byThe Batch

    dlai.onelink.me First shared 2026-09-18
  17. Shared by 1

    only, 18 and over via iOS, Android, muse.ai, WhatsApp

    “Availability: U.S. only, 18 and over via iOS, Android, muse.ai, WhatsApp” via The Batch

    Reported byThe Batch

    pbs.org First shared 2026-09-18
  18. Shared by 1

    Two months later, independent developer Simon Willison identified the “lethal trifecta” for AI agents: private data,…

    “… ETH Zurich researchers led by Edoardo Debenedetti proposed CaMeL, which separates a model that plans from a model that reads untrusted data and enforces written policies before tool calls. Two months later, independent developer Simon Willison identified the “lethal trifecta” for AI agents: private data, untrusted content, and a way to send data out.…” via The Batch

    Reported byThe Batch

    arxiv.org First shared 2026-09-18
  19. Shared by 1

    Willison argued that the only safe option is to avoid combining them.

    “…later, independent developer Simon Willison identified the “lethal trifecta” for AI agents: private data, untrusted content, and a way to send data out. Willison argued that the only safe option is to avoid combining them. Muse processes all three but routes outgoing data through a component the model can’t override, according to Meta. A classifier trained…” via The Batch

    Reported byThe Batch

    simonwillison.net First shared 2026-09-18
  20. Shared by 1

    A classifier trained on past prompt injections may catch 99 percent of new ones, but that remains an unacceptable risk,…

    “… Muse processes all three but routes outgoing data through a component the model can’t override, according to Meta. A classifier trained on past prompt injections may catch 99 percent of new ones, but that remains an unacceptable risk, Willison wrote. Accordingly, Meta built the container, credential separation, and Sentinel to hold when initial layers…” via The Batch

    Reported byThe Batch

    simonwillison.net First shared 2026-09-18

The week to that day two or more sources

  1. Shared by 5

    Claude Cowork and chat are now one Claude

    “Anthropic combined Cowork and Chat into one main Claude app, while also launching its built-in Word and PowerPoint versions called Claude Docs and Slides in beta.” via The Rundown AI

    Reported byThe Rundown AI, Superhuman AI, TLDR AI, Ben's Bites

    Discussed onHacker News 234 pts · 225 comments

    How 1 source put it
    claude.com 3 min read First shared 2026-09-16
  2. Shared by 5

    Introducing System One Models & Jev

    TypeSafe AI launches Jev, a System One model that makes structured decisions 40x to 200x faster than existing LLMs. Includes a pricing table showing $0.042 input costs and free outputs, plus workflow evals comparing Jev against GPT-5.6 Terra. cited

    Reported byThe Rundown AI, TLDR AI, AI Breakfast, Ben's Bites, Superhuman AI

    Where it was shared
    typesafe.ai announcement · 10 min read First shared 2026-09-16
  3. Shared by 9

    Dario Amodei

    Anthropic CEO Dario Amodei calls for slowing AI capability advancement to let safety measures catch up. Details a three-step plan including embedded third-party evaluators, democratic coordination on safety standards, and global government coordination. cited

    Reported byExponential View, TheSequence, The Rundown AI, Superhuman AI, TLDR AI, AI Breakfast, Turing Post, Forward Future, Last Week in AI

    Where it was shared
    darioamodei.com announcement · 17 min read First shared 2026-09-13
  4. Shared by 4

    OpenAI’s new rules for reporting model misbehavior

    “OPENAI📂 OpenAI’s new rules for reporting model misbehaviorImage source: OpenAIThe Rundown: OpenAI just released six new reports on its models misbehaving during training, including wild details like rewriting their own jailbreak-style instructions to using leaked credentials, alongside a new process for making such incidents public faster.The details: An…” via The Rundown AI

    Reported byDon't Worry About the Vase, Ben's Bites, The Rundown AI, Superhuman AI

    How 1 source put it
    • Don't Worry About the Vase2026-09-17issue ↗
    • Ben's Bites2026-09-17issue ↗
    • The Rundown AI2026-09-18issue ↗ OPENAI📂 OpenAI’s new rules for reporting model misbehaviorImage source: OpenAIThe Rundown: OpenAI just released six new reports on its models misbehaving during training, including wild details like rewriting their own jailbreak-style instructions to using leaked credentials, alongside a new process for making such incidents public faster.The details: An…
    • Superhuman AI2026-09-18issue ↗
    openai.com First shared 2026-09-17
  5. Shared by 8

    Countering misuse of AI: September 2026 / Anthropic

    “Anthropic's Threat Intelligence team has identified and disrupted several operations in the past several months where threat actors tried to use Claude for malicious activity. This report shares case studies from those operations and describes how malicious use of Claude has evolved since…” via TLDR AI

    Reported byThe Rundown AI, Superhuman AI, TLDR AI, AI Breakfast, Interconnects, Ben's Bites, Last Week in AI, The Batch

    How 1 source put it
    anthropic.com 156 min read First shared 2026-09-11
  6. Shared by 3

    Toward Recursive Self-Improvement: How GLM Built Its Own Inference Infrastructure

    “Z AI revealed that its GLM-5.3 AI helped set up the 100K-chip system now serving GLM-5.3-Flash, writing that “our successors are the AI systems we are creating ourselves.”” via The Rundown AI

    Reported byThe Rundown AI, TLDR AI

    Discussed onHacker News 411 pts · 285 comments

    How 1 source put it
    • The Rundown AI2026-09-18issue ↗ Z AI revealed that its GLM-5.3 AI helped set up the 100K-chip system now serving GLM-5.3-Flash, writing that “our successors are the AI systems we are creating ourselves.”
    • TLDR AI2026-09-18issue ↗
    z.ai First shared 2026-09-17
  7. Shared by 4

    On the Navier–Stokes Millennium Prize Problem

    “Quartz: OpenAI cracks the Navier–Stokes Millennium Prize problemOpenAI fought dirty on career-making math problem, says NYU mathematicianOpenAI’s feud with mathematicians is only escalating” via Last Week in AI

    Reported bySuperhuman AI, TheSequence, Last Week in AI, The Batch

    How 1 source put it
    openai.com First shared 2026-09-11
  8. Shared by 3

    Meta One is a subscription service that offers enhanced AI and self-expression features across Instagram, Facebook, and…

    “Meta rolled out Meta One, a new paid tier across its apps running from $2.99 to $499 a month for extra Meta AI usage and creator tools.” via The Rundown AI

    Reported byThe Rundown AI, TLDR AI, Ben's Bites

    How 1 source put it
    about.fb.com First shared 2026-09-16
  9. Shared by 2

    Projects redesigned: from folder to conversation

    “🗂️ Projects - Workspace where Claude splits a goal into side-by-side sessions” via The Rundown AI

    Reported byThe Rundown AI, TLDR AI

    How 1 source put it
    claude.com First shared 2026-09-18
  10. Shared by 3

    What’s new: Meta introduced Muse, a personal AI agent based on the Muse Spark 1.3 model.

    “Meta unveiled its first personal agent: Muse connects to your personal accounts to carry out tasks on your behalf, but some users have raised privacy concerns.” via Superhuman AI

    Reported bySuperhuman AI, TheSequence, The Batch

    How 1 source put it
    about.fb.com First shared 2026-09-11