Good morning,
This week I'm going to tell you about a little side-project of mine.
Three years ago I asked "the chatbot who cannot be named, and which I have since banned" the following:
"As a man - what activities should I do, what books should I read, what challenges should I undertake to become a better man and, more importantly, to understand how the world works? Teach me the unknown."
(A note for the women reading: I went the male route because that's my own lens - but if there’s enough interest I’m building something for women. There's a signup for that on the site.)
Anyway.
I got a list. And I followed that list. I read 1984 and Crime and Punishment - but I also watched a congressional hearing on UAPs. I learned about the unseen world.
But it's not all books. I watched high-end TV like Succession and The Wire because it suggested them - and framed them as a learning exercise. Watch Succession for the family dynamics and power structures, not for mere entertainment.
It let me follow courses on personal finance and artificial intelligence. I read books on fatherhood and marriage. I let the machine suggest what would make me grow. I learned how banks work, how cryptography works, how to play chess better.
My list evolved as the LLMs got smarter.
For three years I've been tweaking this thing, and it's been sitting quietly in my Claude Projects folder.
I had the vague idea to help other people with this "system," and over one weekend - with the help of Claude Code - I created a version for the public. Added a payment system (WITH correct VAT invoicing), added gamification, and built a product.
I themed it around the Norse God Odin. The Allfather who wanders the lands looking for wisdom and who sacrificed his right eye in exchange for the wisdom of the runes. Now THAT is powerful..
I absolutely LOVE this thing. It's the first thing I see when I open my browser.
In fact, I love it enough that I don't feel bad at all asking a small one-time $49 for access. I'm quite proud of it.
If you've read this far and it speaks to you, that link is the whole pitch. It would genuinely mean a lot.
And if you'd rather help me build it: I'm taking on 20 founding members who'll use the system for a full 6 months and tell me the unvarnished truth. You get it free - and in return I ask two things: honest feedback as you go, and, if it works for you, your permission to share your story so the next person can decide. First 20 to reply get in. Mail me at [email protected]
Anyway - I built an idea into a real app in a weekend. I see a lot of you building with Claude Code and Codex now. Let's talk about that next.
Welcome to the Blacklynx Brief.
AI News
Anthropic overtakes OpenAI
Anthropic released Claude Opus 4.8 - topping GPT-5.5 and Gemini 3.1 Pro on agentic coding, computer use, and Humanity's Last Exam - and paired it with a $65 billion raise. The round lifts its valuation to $965 billion, passing OpenAI to make it the world's most valuable AI lab. (Anthropic)
Nvidia bets the whole stack on agents
At COMPUTEX, Nvidia centered its launches on the bet that AI agents will be compute's biggest consumers, debuting the RTX Spark PC chip, the Vera "CPU for agents," and the 550B-parameter Nemotron 3 Ultra. The $5 trillion firm is reorganizing its entire lineup around software that barely existed two years ago. (Nvidia)
Trump softens AI security order
President Trump signed an executive order asking AI labs to voluntarily submit frontier models for a 30-day government security review before release, down from a previously planned 90-day mandate. The order rules out any mandatory model licensing, with Trump saying stricter rules would hinder the AI race with China. (The White House)
Hackers hijack Instagram by asking Meta AI
Meta patched a flaw that let attackers seize high-profile Instagram accounts simply by asking its AI support tool to change the linked email and issue reset codes. Compromised accounts reportedly included a dormant Barack Obama profile, Sephora, and a U.S. Space Force official; the exploit ran for months. (404 Media)
SoftBank's €75B European AI bet
SoftBank committed up to €75 billion ($87.5 billion) to build 5 gigawatts of AI data center capacity in France - Europe's largest single AI infrastructure investment. The first phase commits €45 billion to deliver 3.1 GW in the Hauts-de-France region by 2031. (Financial Times)
AI Quick News
Florida's attorney general filed the first state lawsuit against OpenAI and Sam Altman, alleging ChatGPT contributed to mass-shooting plans and self-harm.
Google unveiled Coral Board, a low-power development platform built on its Coral NPUs for on-device AI applications.
Anthropic confidentially filed a draft S-1 with the SEC, advancing its race with OpenAI toward the public markets.
Google doubled Omni image generations for Gemini Ultra subscribers and eased usage-limit complaints with free Flash-Lite prompts.
OpenAI expanded Codex with Sites for publishing hosted apps and websites, plus new role-specific plugins.
Cognition raised $1 billion at a $26 billion valuation, citing more than 10x growth since January on the back of its Devin AI engineer.
OpenAI broke ground on "The Barn", a 1 GW Stargate campus in Michigan promising 2,500 union jobs and $45M in Codex credits for in-state students.
Elon Musk clarified that SpaceX's compute deal with Anthropic runs 180 days rather than the three years cited in the S-1.
OpenAI launched Rosalind Biodefense, opening its biology AI to the U.S. government and vetted partners for pandemic preparedness.
Investor Elad Gil argued that AI is "likely in very early lift off and exponential," pointing to December's model releases as a threshold moment.
Anthropic shipped reliability upgrades to Claude Code, improving MCP stability, error handling, session recovery, and long-context compaction.
Microsoft is reportedly merging GitHub Copilot, chat, Cowork, and Autopilot into a single super app.
Apollo's chief economist published data claiming "zero evidence" of AI-driven job losses, arguing cheaper technology is generating more demand and work.
YouTube rolled out automatic detection and labeling of AI-generated videos and shorts, plus wider deepfake-detection access.
Anthropic widened Project Glasswing to 150 more organizations across 15 countries, granting access to its Claude Mythos Preview model.
OpenAI will pull GPT-5.2 and GPT-5.3-Codex from Codex on June 2, making GPT-5.5 the default for free users.
MiniMax released M3, an open-weight model it says beats GPT-5.5 and Gemini 3.1 Pro on coding while nearing Anthropic's Opus 4.7.
An AI consultant told Axios a client burned through nearly $500 million in a single month after failing to cap employee Claude usage.
Cognition folded its Windsurf IDE into Devin Desktop, a single surface for running agents locally or in the cloud.
Robinhood opened its platform to agents with Agentic Trading and an Agentic Credit Card that let AI execute trades and manage spending.
Alphabet is raising $80 billion through a stock sale to bankroll AI infrastructure, with capital spending set to reach $190 billion this year.
Sam Altman said OpenAI is staffing up a robotics division, envisioning "everyone having a personal robot."
CNN sued Perplexity, alleging its AI reproduces articles "verbatim" and surfaces paywalled content.
Closing Thoughts
That’s it for us this week. Please like and subscribe 🙂
