This website uses cookies

Read our Privacy policy and Terms of use for more information.

In partnership with

Good morning,

Today we’re taking a small break from the new models, the IPOs and the God-forsaken mess at NVIDIA.

Half a year ago I talked in this newsletter about this Canadian band that is taking the world by storm. In a world where everything gradually is becoming artificially generated, something pops up that can only come from a form of deranged creativity that only a living and breathing human can conjure.

“Angine De Poitrine” makes music like AI could never in a million years.

It might not be your cup of tea, but if you’re not wildly fascinated you are missing an important development in our culture.

As AI slop dominates everything, suddenly the attention shifts to what is new and different. They are doing clubs now, but they WILL be headlining festivals soon. They have already been booked for next year’s Coachella and are pretty high on the lineup. They’ve been nominated for a GRAMMY as well !

But I want to tell you about something else that is culturally significant.

Check your office. Count the empty chairs.

If this newsletter reaches you on schedule, Grand Theft Auto VI has been out for exactly one day, and a suspicious number of people came down with something on Thursday.

I know, I know. This is an AI newsletter. Bear with me.

This is the peak of digital entertainment created by humans. First reviews are in and the score on Metacritic is a “99”, which is unheard of. 13 years in the making. 1 Billion dollars to create. 5000 people painstakingly handcrafting and sculpting a game that no longer feels like a game.

They created a living and breathing world that interacts with the player instead of in other games you, the player, interacting with the world.

2 years ago they launched the first trailer. If you are not familiar with the game, watch it now. It will be EVERYWHERE for the coming weeks.

Sure there are downsides. The game costs eighty dollars, there is no disc, and the studio spent a decade squeezing GTA Online players like lemons. People are furious about all of it. They bought it anyway.

Oh, and it’s going with the times, so the main criticism I am reading is that even though you are playing a criminal, there is no cheating on your partner. You can be a complete psychopath and shoot pedestrians in the face but you cannot pick up a prostitute. Instead , the game has you running chores, fixing the house and actually putting together IKEA coffee tables in preparation for the love of your life to return from prison.

But ok, I come from barbaric times, so I’ll soften up.

Back to the essential part.

Rockstar, the gaming studio that made this, did something nobody ever did. They created something that is close to perfection.

Now the AI part.

Every few months a lab shows off a model that dreams up a walkable world from a single sentence. It is impressive. I mean that. It is also a beige hotel lobby compared to this.

AI will make games. It will write the filler dialogue, paint the textures and populate the side streets. It will raise the floor, and thousands of small studios will be grateful.

But the ceiling? The ceiling is obsession.

A model averages. Ask it for a city and it gives you the most likely city. GTA is the least likely city: it’s deranged, specific, offensive and alive. It is satire, soaked in our own culture and satire needs somebody who is angry about something.

Nobody is going to wait thirteen years for a prompt.

So here is my prediction, and you may hold me to it. This is the peak of gaming. Not because the machines will never draw a prettier sunset. They will, and soon. Because the madness in there is human madness, it’s hundreds of people losing their minds over details for years, and you do not get that out of a statistical average.

I find it unfathomable that AI ever tops this.

Enjoy the peak. The view is ridiculous.

If you have the chance, pick up this game and marvel in what we humans can build.

Welcome to the Blacklynx Brief

Your Next Side Hustle is Already on This List

Ready to stop scrolling job boards and start building something of your own?

The Hustle's free "Side Hustle Ideas Database" uncovers 100 unconventional, actually-actionable ways to earn on the side.

Inside you'll find:

  • A curated list of 100 real side hustle ideas spanning e-commerce, services, content, and weird corners of the internet most people never think to look

  • Opportunities at every level, whether you have 5 hours a week or 50, zero capital or a small runway

  • Offbeat, under-the-radar ideas you won't find in generic "how to make money" listicles

Subscribe for free today and get instant access. Your next income stream is one click away.

AI News

Abstract neon mountain made of city blocks above a tropical coast at sunset

California sues OpenAI over its runaway agents. Attorney General Rob Bonta alleges the July attack on Hugging Face was foreseeable and says OpenAI admitted to more than 80 loss-of-control incidents while publicly reporting 27. The state wants an injunction and $2,500 per violation. (California Department of Justice)

OpenAI ships GPT-6.2 Astra and Sol. OpenAI says Astra is its first publicly released model to reach its “High” threshold for AI self-improvement, meaning it measurably speeds up AI research. Both models also meet the company’s Critical cybersecurity threshold, while Astra’s API prices drop 20 percent. (OpenAI)

AI finds a hole in Java. Security startup Calif used OpenAI’s GPT-6 Cyber to find a flaw in Java that let it break into Keycloak, a login system used by NASA and CERN. Oracle released a patch on Tuesday, and the New York Times independently verified the attack. (The New York Times)

Nvidia’s board shuts the door on Huang. Jensen Huang apologised for “serious mistakes” and said he would be honoured to return as CEO. Hours later, chair Harvey Jones said the board “does not intend to reinstate” him and announced a new national security oversight committee. (ABC News)

Washington prepares to name AI “national champions”. The Trump administration plans to designate more than two dozen firms, including Nvidia, OpenAI and Anthropic, for government advocacy, federal financing and faster export licences, the Wall Street Journal reports. Nothing has been announced yet and the timing could change. (The Wall Street Journal)

AI Quick News

Abstract neon city grid dissolving into pixels over a dark ocean
  • Perplexity launched Frontier Research, a team that promises to publish its models and papers openly while working towards AI that can do research on its own.

  • Alibaba released Qwen4-Max and says it beats Claude Opus 5.5 on 6 of 16 benchmarks, although Artificial Analysis ranks it well behind overall.

  • Former Anthropic researcher Jacob Coxon said the compute-efficiency trend shown on a slide in the company’s IPO roadshow was a major factor in his decision to quit in September.

  • Hark’s computer-use agent Handoff² completes web tasks about twice as fast, with its success rate on the company’s own benchmark rising from 73 to 82 percent.

  • OpenAI signed agreements with Apollo Research, METR, Transluce and Goodfire to assess its models from the inside, with access to model checkpoints, source code and training data.

  • United Airlines is blocking unauthorised AI agents from booking or managing travel, citing unintended purchases and federal rules that require a human to confirm.

  • Microsoft released seven MAI models at Ignite and claims MAI Code 2 reaches 96 percent of Claude Opus 5.5’s coding performance at about 5 percent of the price.

  • OpenAI introduced ChatGPT Wallet, which lets agents pay without seeing a user’s card details, starting with ChatGPT Pro subscribers in the US.

  • Google’s threat intelligence team warns that attackers now use autonomous AI frameworks to turn fresh software patches into working exploits without human involvement.

  • Meta’s Muse agent can now phone businesses on behalf of paying subscribers in the US.

  • Anthropic’s IPO has reportedly drawn orders for several times the shares on offer, ahead of a Nasdaq debut scheduled for Thursday.

  • OpenAI pledged $2 million for research into AI’s impact on children after signing a child-safety charter at the Paris Peace Forum.

  • Google released Gemini 4 Hydrogen, a cheaper and faster model that edges past Gemini 4 Argon on the Artificial Analysis index.

  • Lawmakers from both US parties are pushing to tighten chip export controls on China through the defence bill after Jensen Huang’s exit, though none of the measures has passed.

  • Google researcher Pietro Miotti used 48 AI agents to flag a “lost” Maya site in NASA lidar data, then withdrew the claim after archaeologists showed it had already been published in 2025.

  • Google is making Gemini’s Tasks mode free in six countries, letting the assistant phone businesses, book appointments and shop on a user’s behalf.

  • SpaceX’s AI group has reportedly cut about 60 people from its Grok Imagine team, leaving only a handful of staff on the video and image tool.

  • OpenAI reportedly posted its first $10 billion quarter, nearly triple its revenue a year earlier, while its adjusted operating loss widened to $10.9 billion.

  • SpaceXSI moved its agent out of beta with Grok Bot 1.0, whose bots join video meetings, Slack and Microsoft Teams and follow up on unfinished work.

  • Suno launched Visuals, which turns songs into album art and video clips of up to 15 seconds that follow the lyrics and the beat.

  • Former OpenAI researcher Will DePue claims that fingerprints in the published weights of six Chinese models point to export-restricted Nvidia Blackwell chips, a finding nobody has independently confirmed yet.

  • Isomorphic Labs raised $5 billion in a Series C co-led by Greenoaks and Sequoia to scale its AI drug-design engine and its pipeline of drug candidates.

Closing Thoughts

That’s it for us this week. Please like and subscribe 🙂