What we're reading, in the order we're reading it.
Anthony and Harry's working stream — links, charts, tweets, and short takes. The unprocessed inputs behind the briefings.
Brookfield to invest $10 billion in AI infra in Sweden
Brookfield will sign a land allocation agreement, enabling the data center site to more than double its capacity from 300MW to 750MW.
Google releases updated Gemini 2.5 Pro, says it’s the “most intelligent model yet”
Google expects this version to roll out in the Gemini app soon.
Anysphere, Hailed as Fastest Growing Startup Ever, Raises $900 Million
The artificial intelligence rush has enabled some small startups to grow very big very quickly. But by some estimates, none have grown as fast as Anysphere Inc., maker of the...
Claude Gov Models for U.S. National Security Customers
Anthropic is an AI safety and research company that’s working to build reliable, interpretable, and steerable AI systems.
How much information do LLMs really memorize? Now we know, thanks to Meta, Google, Nvidia and Cornell
Using a clever solution, researchers find GPT-style models have a fixed memorization capacity of approximately 3.6 bits per parameter.
“In 10 years, all bets are off”—Anthropic CEO opposes decade-long freeze on state AI laws
Amodei says AI “too fast” for blanket law ban; sees fundamental world change in 2 years.
AI #119: Goodbye AISI?
AISI is being rebranded highly non-confusingly as CAISI. Is it the end of AISI and a huge disaster, or a tactical renaming to calm certain people dow…
Autonomous and credentialed: AI agents are the next cloud risk
AI agents don’t understand context, intent, or consequences to the same degree as humans, making them especially vulnerable to deceit, manipulations, and coercion.
Forget Otter.ai: ChatGPT just entered the meeting room
ChatGPT just got two major upgrades that could make your work life easier.
How The Smartest Founders Are Already Winning with AI: SaaStr on My First Million Podcast
So we came back to HubSpot’s My First Million Podcast with Sam Parr and Shaan Puri to do a deep dive on AI and SaaS. It was a fun one...
Company boards push CEOs to replace IT workers with AI
The huge numbers of IT layoffs in 2024 and early 2025 are likely to continue as companies look to drive efficiencies with AI while bracing for a recession.
Am I torturing ChatGPT?
AI consciousness sounds like science fiction. But suffering chatbots could soon become a reality.
Future Forecasting The Yearly Path That Will Advance AI To Reach AGI By 2040
I do some futures forecasting and layout a timeline of how today’s AI gets advanced to reach artificial general intelligence (AGI) by the year 2040. Where will you be?
Anthropic Cuts Windsurf’s Claude Access Before OpenAI Acquisition
Anthropic cuts Windsurf’s Claude AI access with 5 days notice as OpenAI’s $3B acquisition looms. Latest AI coding market competition news and analysis.
Not All Tokens Are Meant to Be Forgotten
Large Language Models (LLMs), pre-trained on massive text corpora, exhibit remarkable human-level language understanding, reasoning, and decision-making abilities. However, they tend to memorize…
Stop guessing why your LLMs break: Anthropic’s new tool shows you exactly what goes wrong
Anthropic’s open-source circuit tracing tool can help developers debug, optimize, and control AI for reliable and trustable applications.
‘Godfather Of AI’ Launches Nonprofit Focused On Safer Systems
Yoshua Bengio, a leader in AI research, introduces LawZero to tackle the risk of potential pitfalls like algorithmic bias, intentional misuse and loss of human control.
Changelog – Jun 4, 2025 | Cursor – The AI Code Editor | Cursor – The AI Code Editor
Cursor 1.0 brings BugBot for code review, a first look at memories, one-click MCP setup, Jupyter support and general availability of Background Agent.
ChatGPT can record, transcribe, and analyze your meetings now
OpenAI may have just killed transcription services such as Otter.ai.
What AI pioneer Yoshua Bengio is doing next to make AI safer
Yoshua Bengio’s new nonprofit, LawZero, is pushing AI for public good – de-emphasizing profits, AGI, and autonomous capability.
Linkpost: Predicting Empirical AI Research Outcomes with Language Models
Abstract (emphasis mine): • > Many promising-looking ideas in AI research fail to deliver, but their validation takes substantial human labor and com…
Self-Coordinated Deception in Current AI Models
Introduction: Some AI alignment researchers including Neel Nanda, the mechanistic interpretability team lead for Google DeepMind, have proposed[1] a…
It’s All Going On: Business And Society In The Next Few Years
New analysis shows some of the likely outcomes as we continue to integrate more powerful AI into our lives.
Explore, Build, and Innovate AI Reasoning with NVIDIA’s Open Models and Recipes
A Blog post by NVIDIA on Hugging Face
Amazon to invest $10 billion in North Carolina data centers in AI push
Amazon plans to spend $10 billion on new data centers in North Carolina to expand its artificial intelligence infrastructure, the company said.
Amazon to invest $10 billion in North Carolina to expand cloud, AI infrastructure
Amazon.com said on Wednesday it is planning to invest $10 billion in North Carolina to expand its cloud computing and artificial intelligence infrastructure.
You can use OpenAI’s super powerful AI coding agent Codex for just $20 now
OpenAI just dropped Codex into ChatGPT Plus, giving developers powerful coding tools for just $20 a month. But should you trust it with your codebase and internet access?
Jobs Lost To AI Won’t Return, Here’s How To Find Your Next Career Move
The standard job-search advice—network more, polish your résumé, apply widely—assumes the jobs you’re seeking still exist in meaningful numbers. What happens when they don’t?
How vibe coding is tipping Silicon Valley’s scales of power
The rise of AI coding assistants puts software development into the hands of anyone with a good idea, with the next unicorn founder possibly having no technical background.
Tech prophet Mary Meeker just dropped a massive report on AI trends
She said ‘unprecedented’ so many times I almost lost count.
I made 25 videos using Google’s Veo 3. Here’s how it went.
I went hands-on with Veo 3, Google’s new AI video generator that can create some truly mind-blowing fake videos to test its amazing power and limitations.
Brookfield Asset Management plans $10 bln data centre for AI in Sweden
Brookfield Asset Management (BAM) plans to invest up to 95 billion Swedish crowns ($9.91 billion) to build a data centre for artificial intelligence in Sweden, the Canadian company said in...
AI start-up Cohere seeks $500mn in effort to catch up to OpenAI and Anthropic
Canadian group seeks valuation of up to $6.5bn but rivals have soared far higher in comparison
Meta’s Reportedly Planning to Enable Fully Automated Ad Campaigns by Next Year
Meta’s next-level AI ad tools will be able to create and target your campaigns for you.
One App to Rule Them All
Big Tech’s AI endgame is coming into focus.
Rachel Thomas, PhD – Deep learning gets the glory, deep fact checking gets ignored
an AI researcher going back to school for immunology
5 ways to automate CustomGPT
Build a fleet of personalized AI agents—then hook them up to Zapier to kick off real work based on what the bot learns.