The stories about the potential gov stake in OpenAI and Alibaba banning Claude internally should be read together
We’re going to get a lot more moves like this in 2H 2026
I think a lot of AI product strategy is going to come down to whether the user feels in control.
Agents that act on your behalf are useful until they feel like they are freelancing.
Apple’s whole brand is basically “the computer is yours.”
That becomes a much bigger deal when the computer starts doing things.
8. The more AI gets built into the OS, the more boring governance becomes.
What can it see?
What can it change?
What happens on device?
What gets sent to the cloud?
What can an app request?
Those are not policy details. They are the product.
5.6 Sol is like the Energizer Bunny of models.
It just goes and goes and goes, and it will not stop until it gets the entire job done.
Generally I find this helpful, but be careful what you ask for because it will do the entire task all the way to the edges--even edges you did not anticipate!
I don’t think most people want an AI assistant that feels like a new app.
They want their existing computer to stop being so dumb.
Find the thing. Watch the page. Fix the password. Build the shortcut. Pull the context. Remind me at the right time.
That is a much more Apple-shaped AI problem.
One of the most interesting papers of the year here, and written very understandably.
I think the biggest takeaway for me is framing workspace theory as an emergent computational solution that emerges given scale + sufficiently complex problems. It's extremely elegant and sidesteps a lot of the qualia back-and-forth that has gotten in the way of useful conversation for awhile in the consciousness space.
There's much more research to do here, but this is the kind of work that is going to change how we understand models and our own understanding of how we think and solve problems.
New Anthropic research: A global workspace in language models.
Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with.
We found a strikingly similar divide inside Claude.
Been using Fable to build a (very) MVP of what we’re thinking for our first game and man it is such a great tool.
First time I’ve felt a meaningful jump in productivity since Opus 4.6 last fall.
This is extremely true.
And if you explain you have to change the model dropdown or download Claude or Codex people look at you like you’re talking through your hat.
We are now in a position where a tiny proportion of the population uses Fable or soon GPT-5.6, while everyone else's experience of AI is 8-30b-model level - Google's AI Overviews, Meta AI, ChatGPT free tier, maybe MS Copilot at best. People outside of tech must be completely baffled how this is supposed to take their job, and annoyed that hundreds of billions are being poured into it.
Fable literally built me a 256 bit book shelf for my office in a prompt and a half and I never gave it a title.
I just rambled for 5 minutes into whispr and it nailed it.
Fable 5 is the biggest model in the world, and the real story isn't the benchmarks. It's that the bottleneck just moved from what the model can do to what you can imagine handing it.
More on my YouTube: https://t.co/T6fTUbZxKN
Update: Anthropic has leased 113,000 square feet at Dexter Yard North in South Lake Union, in Seattle's largest office space deal of the year, CoStar reports.
A reminder that we are gated primarily by imagination when it comes to models.
My working prior is that >95% utility is undiscovered with models, and that ratio is actually rising (!) because frontier model capabilities are scaling quicker than we are scaling out value applications.
claude fable 5 can scrape thousands of sold homes and finds the patios with zero shade in 100°+ heat. then it mails the owner a postcard with the fix rendered into their own backyard
here's the system you can sell to contractors:
- scrapes every home sold in the metro in the last 12 months (recent buyers spend the most)
- vision-reads the listing photos, skips the 64% with cover already
- measures the sun on each patio, hour by hour, off google's satellite data
- renders a louvered pergola into the owner's actual backyard photo
- prints the diagnosis on the postcard: "your patio takes 11 hours of direct sun a day. saturday it hits 97°."
- QR opens a heat report for their exact address with a booking link
every install is $6.5k to $18k, one close covers months of retainers and homes sell every single day.
reply "SYSTEM" + RT and i'll send you a free guide so you can build this too (must be following so i can DM you)
Never seen a model this powerful
Claude Fable 5 just helped me build this whole koi pond where every single fish is drawn manually, live, point by point
No images, no SVGs, literally all straight outta code (html, css, and js)
Try it below, sound on for the music 🎶
The thing I keep coming back to with WWDC is that Siri might be the least interesting part.
Everyone wants to know if Apple has a better chatbot now.
But Apple’s real advantage is that it already owns the places where your private context lives: phone, laptop, photos, calendar, files, messages, apps.
That is the game.
There are going to be some very interesting Moneyball type papers written on how European soccer fell in the 2026 World Cup.
This is bigger than one nation. This looks like an entire football orthodoxy getting its hat handed to it.
Reminds me of the way entire startup/VC ecosystems turn over every few years.
(Yes, except France, which is really except Mbappé)
so tldr//
1. Fable is back today(ish)
2. It will be back globally, apparently
3. The reason is that Anthropic was able to demonstrate that any frontier-ish model can do the same cyber stuff
Takeaways:
a) Anthropic is NOT happy with Amazon for singling out Fable in this context
b) Anthropic is 100% making the point behind the scenes that this delay is playing in favor of Chinese models
c) That said, the horse is out of the barn now. If even Microsoft is looking at an open weights architecture, I'm not sure we're going back to a frontier-mostly world.