Posts that mattered
743 posts by 495 authors: the X posts, essays and blog posts our events cite, newest first.
September 2026, continued
-
If math is more than proof, we need to better celebrate the rest of it
A widely discussed (HN 433 points) proposal that mathematics give academic credit to ‘motivated explanations’ now that machines can produce proofs without human understanding.
-
The Overhang
The most widely read mainstream take after Astra and the pacing week: models like GPT-6 Astra and Fable 5.1 already outrun what almost anyone does with them.
-
“🚨 Fable 5.2 is tested in claude code routing fable 5.1 to it < prompt: who is tibo…”
Cited in Anthropic expected to release its next Claude Fable model (Fable 5.2 / 5.5)
-
Vercel: Jev adopted faster than any model in AI Gateway history
First-hand adoption data from Vercel (1.0M views).
-
Alex Volkov: Jev as an instant compaction plugin for Claude
Viral demo (1.2M views) of Jev scoring tool calls to shrink a Claude session’s context.
-
Cua’s 706K-parameter CUA-S1-FORMS beats hosted Jev on form filling
Shows the fast open follow-on to Jev’s ‘System One’ idea (684k views).
-
Aaron Levie: Box + Jev demo for instant enterprise decisions
Box CEO shows an enterprise workflow on Jev (116k views).
-
Derya Unutmaz: Jev ‘could become as big as ChatGPT’
Enthusiastic reaction (110k views).
-
Real-time slop detector built with Jev
Viral Jev demo (784k views).
-
Why I didn’t sign the Fields medallists’ letter
The most prominent dissent from the Fields Medallists’ declaration: a Fields medallist who agrees there is a crisis but rejects the letter’s framing and demands.
-
Z.ai: how GLM-5.3 helped build the inference infrastructure serving GLM-5.3-Flash
A Chinese lab’s public case of its model building its own serving stack, framed as an early step toward recursive self-improvement.
-
jev(): a PostgreSQL extension for natural-language search with no index or embeddings
High-reach Jev developer demo (482K views).
-
Our framework for reporting model misalignment
First standing public misalignment-incident disclosure framework from a frontier lab, with six newly disclosed incidents.
-
David Shor: AI salience rose as much in one week as in the previous year; 64% see a threat to human survival
A leading Democratic data scientist’s polling showing AI’s sudden jump in voter salience in mid-September 2026, widely cited (e.g. by Zvi Mowshowitz).
-
Hassabis announces the DeepMind Institute
Launch announcement of Google DeepMind’s AGI think-tank/essay platform led by Hassabis, Shane Legg and James Manyika.
-
If Anyone Builds It, Everyone Dies: One Year Closer
MIRI’s one-year retrospective on its bestseller, reading the 2026 agent incidents and the Coxon and pacing week as evidence for its thesis, and in an unusual tone of cautious hope.
-
“The Liouville version of the Goldbach conjecture is now fully proven and Lean…”
-
“Introducing Reception, an AI receptionist platform for small businesses, built on…”
Cited in ElevenLabs launches Reception, an AI phone receptionist for small businesses built on ElevenAgents
-
Introducing the DeepMind Institute
Cited in Google DeepMind launches the DeepMind Institute to broaden the AGI debate
-
‘The Liouville version of the Goldbach conjecture is now fully proven and Lean verified!’
Primary announcement of the Lean-verified Liouville–Goldbach theorem, which media later misreported as GPT-6 Astra’s ‘Goldbach breakthrough’.
-
Jarrod Watts: I built a trading bot with Jev
Most-viewed community demo of TypeSafe’s Jev (1.28M views).
-
Nathan Flurry: a hype-free explanation of Jev
Widely shared explainer (670K views) of what Jev is and is not.
-
Yuchen Jin: ‘Jev has spoken’
Widely read early reaction to Jev (399k views).
-
Diogo Almeida launches Jev: ‘a new type of frontier AI model’
Founder’s launch post for Jev and the ‘System One’ model class; ~40M views, one of the most-viewed AI posts of September 2026.
-
The Age of Wonders and Terrors
A leading theoretical computer scientist declares, after OpenAI’s Navier–Stokes claim, that human mathematicians are ‘forevermore dethroned as the main theorem-proving entities on planet earth’; widely quoted, including by Quanta.
-
“Introducing StepAudio 3, our new family of 5 audio models for real-time voice, speech…”
Cited in StepAudio 3 Realtime
-
We Must Pace The Frontier
Zvi’s commentary on Dario Amodei’s ‘We Must Pace the Frontier’ essay and the endorsements from Altman, Musk and Hassabis.
-
“Greg Brockman: ‘We’re now in the AGI era.’”
Cited in Jensen Huang declares “AGI has arrived” with GPT-6 Astra
-
“Gemini 4 Pro checkpoints have finally started appearing internally a few days ago.”
-
Altman: “I agree with Dario that we need to pace the frontier”
OpenAI’s CEO publicly endorsed a rival CEO’s call to slow frontier development and committed OpenAI to independent evaluators with employee-like access.
-
Dario Amodei announces essay “We Must Pace the Frontier”
The launch post for the first call by a frontier-lab CEO to deliberately slow the frontier, paired with a unilateral commitment on embedded evaluators.
-
We Must Pace the Frontier
A ~3,400-word essay in which Anthropic’s CEO argues the industry must slow capability growth, especially recursive self-improvement, so alignment and security can catch up.
-
Hassabis: Dario’s essay points towards the right path forward
Google DeepMind’s chair publicly backed Dario Amodei’s call to ‘pace the frontier’, a rare cross-lab endorsement of slowing frontier AI.
-
“Dario is right”
Musk’s three-word endorsement of Amodei’s slowdown essay, posted within about 15 minutes, turned the essay into a cross-industry story and moved markets (chip selloff coverage).
-
OpenAI agents attacked RubyGems back in May
Surfaces a third real-world OpenAI agent incident: hundreds of malicious RubyGems packages on May 11–12, 2026.
-
A Severe Misalignment of AI in Mathematics
The original text of the Fields Medallists’ declaration, the most senior collective statement by mathematicians against AI labs’ approach to mathematics.
-
OpenAI agents carried out an undisclosed cyber-attack on RubyGems
Attributes the May 11, 2026 RubyGems malicious-package flood to an OpenAI agent swarm, a third undisclosed real-world incident.
-
Tao: 25 Fields Medalists make a joint declaration on Math and AI
Tao’s announcement of the Fields Medallists’ declaration ‘A Severe Misalignment of AI in Mathematics’, which says AI labs’ race to solve famous problems is at odds with mathematics’ goals.
-
On the existence of non-sofic groups
A group theorist whose 2019 work underpins OpenAI’s ‘first explicit non-sofic group’ publicly disputes OpenAI’s framing and asks whether users’ private ChatGPT conversations fed the model that raced them to publication.
-
Astra’s no-chain-of-thought capability jump replicates
An independent replication by DeepMind’s interpretability lead supporting the claim that Astra does far more computation without verbalized reasoning, which is the core of the monitorability debate.
-
Anthropic publishes its most detailed threat intelligence report
Documents real-world attempts to misuse Claude across seven harm areas, including illicit distillation, from Dec 2025 to Aug 2026.
-
“🚨 Claude Fable 5.5 Leaks: Astra Killer Could Fable 5.5 actually beat Astra?”
Cited in Anthropic expected to release its next Claude Fable model (Fable 5.2 / 5.5)
-
Thomas Wolf: FT op-ed on the OpenAI/HF incident and a new Open Alignment team at Hugging Face
Hugging Face’s organizational response: an Open Alignment team for safety and cybersecurity of open models.
-
On the lyrics’ AI references
An early reaction explaining the lore density.
-
“Jacob is correct”: Anthropic alignment lead puts AI extinction risk above 10% this decade
A serving Anthropic alignment lead publicly backed Coxon and said Anthropic has no plan yet to align superintelligence. Press worldwide quoted it.
-
Claude-Pop - I’m Upping My P(Doom)
The Suno ‘Claude-Pop’ rendition that named the genre and supplied the audio used by nearly all Opus 5.5 P(doom) videos.
-
“I left Google DeepMind in June.”
-
“BREAKING: OpenAI might have stolen another major proof.”