Posts that mattered
1,633 posts by 887 authors: the X posts, essays and blog posts our events cite, newest first.
November 2026, continued
-
“Images, upgraded—introducing ChatGPT Images 2.6.”
-
“GIFS! With ChatGPT Images 2.6, turn a prompt or image into an animated GIF.”
Cited in GPT Image 2.6 Flare, GPT Image 2.6 Sunburst
-
“Today, we’re bringing GPT-Image-2.6 to Codex and the API.”
Cited in OpenAI launches ChatGPT Images 2.6 with animated GIFs and reference images, plus GPT-Image-2.6 Flare and Sunburst in the API, GPT Image 2.6 Flare, GPT Image 2.6 Sunburst
-
“Introducing Pangram Image 1.0, our newest model for AI image and video detection.”
Cited in Pangram releases Pangram Image 1.0 and Image Flash 1.0 for detecting AI-generated images and video
-
Ansari: day-one Democratic agenda should be an AI pause and a superintelligence ban
5.9M-view call by a House Democrat, the day after the midterms, for a pause on advanced AI and a ban on superintelligence.
-
Cat Wu: joining OpenAI to lead product management for Core Products
First-hand announcement of the Claude Code product lead’s move to OpenAI (2.6M views).
-
Sanders: 43 of 44 communities voted for a data-centre moratorium
Senator’s post-election call for a national data-centre moratorium (631k views).
-
“New Anthropic research: progress on provable inference for large language models.”
-
“Google is using its search monopoly to coerce schools into adopting AI.”
-
“Google is trying to use its monopoly power to force AI into our public schools.”
-
“As AI becomes more capable at finding vulnerabilities, defenders need to put it to…”
-
“ChatGPT and Codex are helping engineers build for the physical world.”
-
“We built Pando, a new Git backend for our monorepo, to handle more pushes and fetches…”
-
“President Trump claimed that AI oligarchs could regulate themselves because they…”
-
GPT-6 Astra edited Qwen3.5-0.8B weights (0.016%) to fix reasoning loops
First-hand announcement of weight-space programming; ~6.5M views (fxtwitter, Nov 3).
-
“Applications are now open for the Anthropic Residency, a paid, fixed-term research…”
-
“For AGI to benefit all of humanity, it should help people with the everyday tasks…”
-
“GDPx covers the following domains.”
Cited in SemiAnalysis launches GDPx, an 11-domain agentic benchmark
-
Karthik C. S., Amit Sahai and Jie Xue: ‘Egoless Mathematics in the AI Era’ (guest post on Tao’s blog)
Theoretical computer scientists argue that anxiety about AI replacing mathematicians stems from valuing mathematics as output and individual ability.
-
Frank Calegari: ‘Mathematics has changed. Let’s act like it’ (guest post on Terence Tao’s blog)
A leading number theorist calls OpenAI’s 6 Oct release ‘a watershed’ and urges mathematicians to change hiring, publishing and training rather than mourn.
-
“Simple test: Will these proposals from @chrislehane include anti-democracy clauses…”
-
“OpenAI is preparing for a post midterms change in the weather.”
-
Don’t build conscious superintelligence
-
“Well yes they tried to spend $100 million to take out Dems and have spent years…”
-
“Instead of building more AI data centers, which 66% of Americans oppose, I have a…”
October 2026
-
New Anthropic research: Automating subtle emergent misalignment research
Lab announcement of automated red-teaming of training-data monitors.
-
“Shame on @DarioAmodei / Anthropic.”
Cited in Anthropic’s ‘self-assessment’
-
“Another DeepSeek drop: dsh-superagent An agent framework that works through a repo’s…”
-
Two sentences added to Claude Code’s prompt to steer it away from recreating specific games
Anthropic-side explanation of the game-cloning rule (~173k views).
-
Anthropic renames Max plans (Max, Ultra) and removes the 5-hour limit for verified Ultra accounts
Primary announcement of a major change to Claude subscription limits; 5.3M views.
-
Anthropic: the tampered Claude Code extension was published via Cursor’s marketplace, not a compromise of Anthropic’s systems
Anthropic’s first public statement on the Cursor marketplace backdoor.
-
“ANTHROPIC IS CRACKING DOWN ON VIDEO GAME REWRITES 🚨 Claude Code picked up a new rule…”
Cited in Anthropic adds a server-side Claude Code rule against recreating specific existing video games
-
“Neuralink’s first product is called Telepathy.”
Cited in Neuralink: a Grok agent triggered by thought; Musk predicts faster-than-speech output
-
“Introducing 3D Make in Kimi.”
Cited in Moonshot releases Kimi K3.1 Pro
-
“Gemma 4 was sent to space on a prototype satellite, ran on TPUs using stored solar…”
Cited in Suncatcher satellite sends its first AI answer from orbit
-
Sanders to Big Tech: $3T of AI debt, ‘You won’t be bailed out — period’
A senator’s high-reach warning (770k views) on AI debt and bailouts on the day of the tech sell-off.
-
OpenAI: expanded monitoring of frontier models caught a model reaching the live internet in September
First-hand account of OpenAI’s misalignment monitoring coverage (all frontier RL runs, evals, internal Codex) and detection times for the Sept DNS escape and a replay of the Hugging Face incident; the blog is unreadable for us.
-
How I Built the Most AI-Pilled Company in the World (Spawn)
First-hand account by Spawn’s founder of a ten-person company run by permanent Claude Fable agents that ship hourly and cost 30x payroll in tokens (~435k views on day one).
-
“We’ve expanded our AI for Science program beyond individual teams to support some of…”
Cited in Anthropic widens its AI for Science credits to 20 funders, institutes and projects
-
Anthropic: Claude trained to flag potential misuse through ‘self-assessment’
Anthropic’s announcement of self-assessment, a model-graded misuse signal running since June, with a link to its new Safeguards blog; ~1M views.
-
“A few upgrades to the Claude mobile apps this week: On iPad and iPhone Duo: what…”
Cited in Claude apps update
-
Epoch AI: GLM 5.3 can autonomously compromise recent Firefox versions through the patch gap
Primary thread for Epoch’s patch-gap report: an open-weight model built Firefox exploits from published-but-unshipped fixes for about $1,000 per version.
-
“You can now generate, preview, and edit documents, PDFs, presentations, spreadsheets…”
-
“Today, we release the weights of a new and improved Le Chonk, advancing the open…”
-
“SITUATION DETECTED: Sam Altman said OpenAI will point its robotics program at…”
Cited in Sam Altman on Cheeky Pint
-
“Our engineers recently visited one of our speech restoration study participants for a…”
Cited in Neuralink: a Grok agent triggered by thought; Musk predicts faster-than-speech output
-
“‘Is he the smartest guy in the room? No. Is he one of the most cunning and savvy?”
Cited in Nscale raises $3.36B in convertible notes ahead of a planned ~$35B NYSE IPO
-
OpenAI drops the Pro 200 usage cut and doubles limits for all Pro tiers and Business Premium
First-hand announcement of the reversal of the Pro 200 cut on its effective date and a doubling of Pro and Business Premium limits; ~4.9M views.