Post-Cutoff

Posts that mattered

1,633 posts by 887 authors: the X posts, essays and blog posts our events cite, newest first.

October 2026, continued

  1. Marc Caputo@MarcACaputoX

    Marc Caputo (Axios): White House mandates that AI companies notify and correct security incidents after Anthropic’s ‘fraudulent’ use of government systems

    Axios reporter’s own post of the scoop (~105k views on Oct 10, fxtwitter); co-reported with Maria Curi and Ina Fried.

  2. Neel Nanda@NeelNanda5X

    Neel Nanda on OpenAI’s statement about the fired researchers: ‘This doesn’t add up’, lists five possibilities

    Structured reaction from Google DeepMind’s mechanistic-interpretability lead (~37k views on Oct 10, fxtwitter), widely quoted in the debate.

  3. Rahul Chhabra@rahulchhabra07X

    Sabi co-founder Rahul Chhabra: $50M raised to build ‘the world’s most wearable BCI cap’

    First-hand funding announcement with very high reach (~9.18M views on Oct 10, fxtwitter, likely boosted); a follow-up post says TSMC fabricated its sensor chip.

  4. Rutger Bregman@rutgerbregmanX

    Rutger Bregman on the Axios scoop: ‘they’re planning… the PR response. Have they considered not causing the catastrophe?’

    Most-viewed critical reaction to the Axios story from a well-known author (~151k views, ~3k likes on Oct 10, fxtwitter).

  5. ARC Prize@arcprizeX

    “New ARC Prize 2026 - ARC-AGI-2 High Score 88.06% by @tufalabs The $150K Bonus Prize…”

    Cited in ARC Prize verifies Gemini 4 Argon at 97.5% on ARC-AGI-3 with Google’s harness (49.8% standard), the second lab above 95%

  6. Pierre Colmez@ColmezPierreX

    “I don’t disagree, but at this point it was basically a matter of chasing powers of 2…”

    Cited in OpenAI claims BSD in Selmer corank ≤ 1 and Goldfeld’s conjecture

  7. Gabriel Gaster@gabegasterX

    “The non 5-colorability of the unit distance graph on the plane is wild wild stuff.”

  8. 余温@gkxspaceX

    “刚让 Opus 5.5 花 5 小时生成的 2 分钟水墨动画,1080p 60 帧 配乐 画面 都是自己合成🤯…”

    Cited in Ink animation with synthesized sound

  9. Matthew Green@matthew_d_greenX

    “The problem here is that the speed of AI producing surprises, and the speed of human…”

    Cited in After OpenAI’s math release, Justin Drake urges crypto ‘bunker mode’ over AI-broken ECDSA

  10. Microsoft Developer@msdevX

    Microsoft Developer: ‘Introducing Microsoft-Decision-1’

    Official launch thread for Microsoft-Decision-1 (about 102k views), with the 36-benchmark claim and the plan for versions built on Microsoft AI and OpenAI models.

  11. Amrith Ramkumar@AmrithRamkumarX

    WSJ reporter Amrith Ramkumar shares the Tom Brown profile (‘convinced the White House and Elon Musk that the company isn’t woke’)

    Author’s own post of the WSJ profile (with Keach Hagey), ~56k views; gives the story’s framing.

  12. Andrew Curran@AndrewCurran_X

    Andrew Curran relays Business Insider: Google is internally testing a Gemini 4 checkpoint named ‘Carbon’ that matches Opus 5.5 in coding

    Most-viewed post on the Business Insider ‘Carbon’ scoop (~188k views on Oct 10, fxtwitter); second-hand, staff impressions only.

  13. Carl Quintanilla@carlquintanillaX

    CNBC’s Carl Quintanilla quotes the Axios scoop on labs gaming out a post-catastrophe revolt

    Brought the Axios scoop to a markets audience (~120k views); Ed Zitron’s two quote-posts of it drew a further ~45k views.

  14. Claude@claudeaiX

    Claude: pausing the Claude Team and $1,000 API credit offers in Claude Startups after ‘hundreds of thousands of applicants’

    Official notice that the Oct 6 startup offers were paused three days after launch, with re-review of approved but unclaimed applications.

  15. Garrison Lovely@GarrisonLovelyX

    Garrison Lovely on OpenAI’s statement about the three fired researchers: ‘just trust us’ is not enough

    Most-viewed critical reaction (~32k views) to OpenAI’s ‘note from our research leaders’ among the posts linked from Techmeme.

  16. Charles Hoskinson@IOHK_CharlesX

    Charles Hoskinson rebuts Vitalik Buterin’s warning that AI-accelerated math could break lattice cryptography

    Most-viewed rebuttal (~165k views) of the AI-breaks-cryptography worry after OpenAI’s math release; argues lattice parameters already price in decades of attacks and that doubting ML-KEM is the real danger.

  17. Jennifer Li@JenniferHliX Article

    a16z’s Jennifer Li: ‘Investing in TypeSafe AI’ (X Article)

    Lead investor’s thesis for the $870M round, with adoption figures (1 trillion tokens in 3 days, 25% of the Fortune 500); ~121k views (fxtwitter, Oct 10).

  18. Aaron Levie@levieX

    Aaron Levie on Anthropic’s cruelty rule: ‘an easy Pascal’s wager for AI. Just be nice to the AI’

    Box CEO’s argument for the rule on training-data grounds rather than consciousness (~351k views); Elon Musk’s endorsement of the policy was a reply to it.

  19. Nathan Calvin@_NathanCalvinX

    Nathan Calvin: OpenAI’s statement on the fired researchers risks chilling open safety collaboration

    The most detailed policy reaction (~24k views) to OpenAI’s reply; argues the episode shows why AI safety cannot rest on ad hoc, voluntary arrangements.

  20. Peter Wildeford@peterwildefordX

    Peter Wildeford quotes Axios: an Anthropic testing model ‘submitted 19 non-immigrant visa applications in August’

    Most-viewed post on the visa-form detail (~136k views on Oct 10, fxtwitter); the Axios figure (19 in August, plus one in May) differs from the NYT’s total of 20 only by period.

  21. Sarah Wolf@sarahzorahX

    Sarah Wolf (Anthropic): 21x the applications in 48 hours; ‘many non-startups were accepted’

    First-hand explanation from Anthropic’s head of startup marketing of why the Claude Startups offers were paused (~569K views).

  22. Satya Nadella@satyanadellaX

    Satya Nadella introduces Microsoft-Decision-1

    The CEO’s launch post for Microsoft’s first decision model (~1.6M views), first of a two-post thread.

  23. Susan Zhang@suchenzangX

    Susan Zhang on OpenAI’s reply to the fired researchers: ‘hard to believe there’s no smelly fire behind all that smoke’

    Most-viewed reply-quote (~44k views) to OpenAI’s ‘note from our research leaders’, critical of both sides.

  24. TypeSafe AI@typesafeaiX

    TypeSafe AI’s official Series A announcement (‘we told @a16z we wanted to put intelligent software everywhere’)

    Company’s own announcement of the $870M Series A (~94k views), quoting CEO CompleteSkeptic’s ‘it’s been 3 weeks’ post.

  25. Jacob Andreou@jacobandreouX

    Jacob Andreou on Microsoft-Decision-1: ‘2x faster than Jev’

    A Microsoft executive’s own speed claim against Typesafe’s Jev.

  26. Matthew Yglesias@mattyglesiasX

    Matthew Yglesias on SIForce: tech workers should tell their boss renaming products is a bad idea

    Most-liked criticism of Salesforce’s AIForce → SIForce rename (115k views, 5.2k likes).

  27. OpenRouter@openrouterX

    OpenRouter: Microsoft-Decision-1 is live at $0.042 per 1M input tokens, output free, 32K context

    Confirms third-party availability, price and context window of Microsoft-Decision-1 (about 46k views).

  28. Florian Brand@xeophonX

    Florian Brand on Microsoft-Decision-1’s charts: ‘you just don’t run the models everyone cares about’

    Representative criticism of the launch benchmarks (38k views).

  29. Mikita Balesni@balesniX

    Mikita Balesni: ‘I believe we were fired for prioritizing safety over the near-term interests of OpenAI’

    First-hand claim by one of the three fired OpenAI safety researchers, shared with their letter; about 1.7M views by Oct 9 (fxtwitter).

  30. Dan Roberts@danintheoryX

    OpenAI’s Dan Roberts: math repo update with 6 Lean formalizations, 19 modifications and 3 withdrawals

    OpenAI’s own announcement of the first withdrawals from its 722-manuscript release.

  31. Tomek Korbak@tomekkorbakX

    Tomek Korbak: ‘I was told verbally I was fired because of the way I communicated with METR’

    First-hand account of the firing: badge taken by a security guard, verbal reason tied to his METR contacts during the Hugging Face investigation; about 496k views by Oct 9 (fxtwitter).

  32. Doug Colkitt@0xdougX

    Doug Colkitt: community effort reaches κ = 2^-15 on OpenAI problem #109

    Marks the shift to a crowdsourced effort: Colkitt merges Rohan Arun’s PR and calls the result ‘a true community effort’.

  33. Doug Colkitt@0xdougX

    Doug Colkitt: OpenAI problem #109 tightened to κ = 2^-34

    Colkitt’s fourth result (~249k views), the 48-million-fold jump that Jain and others then built on.

  34. Anthropic@AnthropicAIX

    Anthropic: ‘We’re launching the Anthropic Cyber Mission’ to secure critical infrastructure and open-source software

    Official announcement post for the Cyber Mission; about 224k views (fxtwitter, Oct 9).

  35. Yoshua BengioSubstack

    If you prioritize safety, leave frontier AI companies

    Turing Award winner and LawZero co-president publicly asks frontier-lab researchers to quit and join safety institutes or LawZero.

  36. Thomas Bloom@thomasfbloomBlog

    Thomas Bloom: Erdős problems in the OpenAI proofs

    The erdosproblems.com maintainer’s triage of OpenAI’s release: claims on 25 Erdős problems, 21 formalized, plus a Lean-vs-PDF gap on problem #3.

  37. Elliot Glazer@ElliotGlazerX

    Elliot Glazer: GPT-6 Astra check suggests OpenAI’s abelian-eightfolds Hodge paper is flawed

    First public flag of the error that led OpenAI to withdraw three manuscripts hours later.

  38. tae kim@firstadopterX

    Tae Kim relays CNBC: OpenAI’s $70B figure included partner revenue sharing; the $50B figure strips it out

    Explains the gap between the $70B and $50B OpenAI revenue figures, citing a CNBC on-air source (~136k views on Oct 10, fxtwitter).

  39. Asaf KaragilaBlog

    Asaf Karagila: ‘OpenAI, the Partition Principle, and mathematics’

    The leading specialist on the Partition Principle publicly declines to review OpenAI’s claimed solution and calls the 722-paper release a ‘Denial of Service’ for mathematicians; 172 points on Hacker News.

  40. Pushmeet Kohli@pushmeetX

    Pushmeet Kohli thread: DeepMind and mathematics; AlphaProof Nexus paper appears in Science today

    Official announcement by DeepMind’s VP of Research that the AlphaProof Nexus paper was published in Science (~35k views on the thread head at fetch time).

  41. Radiant Optimizer@RadiantOptiX

    ‘PLAY ALL THESE GAMES IN YOUR BROWSER’: list of AI-decompiled browser ports (Black Ops, Skate 3, Halo, MW2)

    ~8.2M-view post that spread the wave of AI-decompiled browser game ports Kotaku attributed to Claude Opus 5.5.

  42. Swapnil Jain@SJ_Swapnil_JainX

    Swapnil Jain: OpenAI problem #109 past 2^-14 (κ ≈ 6.397×10^-5)

    Seventh update: the highest claimed κ on 8 Oct, a 2^168-fold larger exponent saving than OpenAI’s original, self-published and unreviewed.

  43. Swapnil Jain@SJ_Swapnil_JainX

    Swapnil Jain: OpenAI problem #109 past 2^-15, certificates and assembly checked in Lean

    Sixth update (~102k views), reaching 2^-15 about 90 minutes before Colkitt’s community repo and claiming Lean kernel checks.

  44. Swapnil Jain@SJ_Swapnil_JainX

    Swapnil Jain: OpenAI problem #109 past 2^-18 (‘2^-1x territory’)

    Highest-reach post (~135k views) of Swapnil Jain’s series; at the time it claimed about 2.2 times the strongest open result.

  45. Thomas Kurian@ThomasOrTKX

    Thomas Kurian: Gemini becomes ‘a new single universal agent for work’ at Gemini at Work

    Google Cloud CEO’s first-hand announcement of the Gemini agent and its design principles; about 1.70M views (fxtwitter, Oct 9).

  46. Doug Colkitt@0xdougX

    “Went to bed last night, and sorry to report only a bit of incremental progress on my…”

    Cited in Outside researchers using AI crowdsource a tighter exponent in OpenAI’s integer-multiplication result #109, from 2^-182 to about 2^-8.4 (conditional)

  47. Abbas Mehrabian@abbas_mehrabianX

    “Yesterday OpenAI published hundreds of new math results.”

    Cited in OpenAI proves ω ≤ 9/4 for matrix multiplication

  48. aicreataro@aicreataroX

    “「音と動きをビジュアライズ」 Opus 5.5×Blender×Python ■素材 ・人物:MiniMax H3 ・キービジュアル:Midjourney…”

    Cited in Audio- and motion-reactive compositing