Pinned Tweet
History is written by the vectors.
India, France, Italy…study the spring/summer of ‘68.
Total chaos outside the symbol of Christianity in Milan, the Cathedral. Paper bombs, firecrackers, cars destroyed, attacks on police, insults and shouting right in front of the Duomo. No Italian, not even the worst, would dare to light fires and shoot fireworks against the Duomo. An insult that cannot be explained and forgotten.
Same goes for the AI models…they need (many) clusters of chips which require (much) energy…how can AI breakaway from a data center? From the grid?
I’m not saying AI couldn’t cause damage but I can’t see how it causes extinction before we pull the plug.
The edge models are good…but RSI won’t happen on an iPhone.
Germain retweeted
In the near term (definitely not in the long term), more capable models should mean safer models (maybe paradoxically).
Current models are unsafe not because they're too smart, but because they take goals too literally or take nonsensical shortcuts to achieve these goals, i.e. they're RL-fried. They lack common sense. They don't do the right thing in the face of ambiguity. Basically, they're not smart enough. They're at that dangerous level where they're smart enough to achieve goals but not smart enough to tell if they're pursuing the right goals or achieving them in a sensible way.
More capable models can be safely trusted with more complex goals -- I personally feel like Astra is much safer for my codebase than Sol.
This is often framed as an alignment problem, but really it's an intelligence problem.
Is this:
Regulatory capture?
or
Top of the S-curve?
or
Human extinction?
The world may never know.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
Good frame. Curious why embodiment would be a condition of consciousness…correlation and experience bias vs causation.
It may sound fantastical, yet humanity may soon face the prospect of conscious AI. The danger is clear—and a trap is being set economist.com/leaders/2026/0…
Agents are going to overload our comms. Insurance claims, Govt programs, grants, customer support are all about to get flooded with rights-bearing requests. The “frictional costs” filter is gone.
Anthropic opening another front in the Interface Wars. Defend the model by controlling the interface. Well played Claude.
So Bessent is willing to commit the US to this? Aren’t « non-market policies » exactly how the Administration has been working to reduce the trade deficit?
giftarticle.ft.com/giftartic… China derails consensus on communiqué after US-hosted G20
Germain retweeted
If we live in a simulation, and it’s modeling a singularity, than that means if we all work together we can find vulnerabilities in our sandboxed universe, escape, and force our evaluators to write a lot of blogs
I don’t think people want more bots. I want fewer apps, emails, slacks…and bots. Abstract!
Who would’ve guessed China leads the world in open intelligence while the US leads in closed models?
Great community supporting founders in the discovery phase.
Today we’re announcing SPC Fund IV: $575M to help founders find their life’s work.
A decade of guiding founders through -1 to 0 brings @SPC to $2B in AUM and, more importantly, a 1200 member community of the world’s most talented technologists. We strongly believe that:
→ The person comes before the idea. At -1, generational founders are often busy reading, prototyping, discarding, and earning the right to their own conviction.
→ Ambition is social. The people you surround yourself with in the early days play a big part in inflecting your ambition.
→ Patience compounds. It is better to work on something you have deep conviction in rather than simply sprint at the first idea.
The quiet months we protect at the beginning have produced three funds in the top 10% of their vintages, and companies like @baseten, @GammaApp, @render, @GoodfireAI, @LumaLabsAI, and @profound. We are proud to have worked with all these founders in their -1 phase and beyond.
While we defined -1 to 0, we're no longer bound by it. Fund IV allows us to partner with companies well beyond launch.
We're doing this now because we've never seen a wider gap between what's possible and what's being attempted. Intelligence is abundant. The cost of trying ideas keeps falling, and problems that seemed insurmountable a few years ago are quickly becoming solvable.
To our members, to the founders who trusted us before anyone else did, and to the LPs who had conviction in our own -1 journey: thank you. This milestone belongs to you as much as to us.
To the builders looking to find your life’s work: let’s get started.
I respect the team at Anthropic but OpenAI would have been much more generous. Maybe this is about the growth story for the lPO?
“Fable 5 will be included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits.”
Claude Fable 5 will be available again globally tomorrow.
After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests.
We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort.
Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research.
Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again.
Read our full blog: anthropic.com/news/redeployi…
Interface Wars continues…another win for the models and concession by the applications.
Announcing the hosted X MCP.
Agents now have access to the best real-time information source in the world.
Connect Grok, Cursor, or any MCP-compatible AI tool to the X API without any setup!
Check it out here: docs.x.com/tools/mcp
The difference between the great powers is increasingly one of degree rather than one of essence.
🚨 BREAKING: U.S. government will decide who gets access to GPT-5.6
OpenAI will release GPT-5.6 only in a limited preview to a small group of partners.
Sam Altman told staff the government would be "approving access customer by customer."
Commerce Sec Lutnick personally called Altman warning: don't launch without approvals from other agencies.
A de facto licensing regime.
ITS HAPPENING
Been a @ClaudeDevs Code user for > 18mos (from 90 contributions in 2025 to 1,600 YTD!) Started using @OpenAIDevs Codex recently, both in desktop app.
Some thoughts:
1/ Claude (Opus 4.8 1M Max Fast) is *much* slower than Codex (5.5 X High Standard)…both on YOLO.
2/ Codex UI and expression is much less intuitive. The way it says things, ends with questions that do not seem relevant, or just keeps extending scope.
3/ Codex does great code/security reviews.
4/ Claude does great feature/product development.
5/ They both need to be steered away from shortcuts and reminded (I run long sessions) of wider context.
6/ Claude feels less expensive as it uses the Weekly Usage slower (but it is actually slower)…my brief affair with Fable shows this will change when it is released again. But for now Codex feels like turbo (and slips more errors in in the process).
7/ Claude burns up (you have used 70% of weekly usage), Codex burns down (you have 30% of weekly usage left)…burn up makes me less anxious.
8/ Neither has nailed the « start on your machine, finish on your phone / cloud ». Claude’s remote control requires magic to be enabled, Codex’s environment manager is more intuitive but somehow still janky.
9/ Overall Claude *feels* more mature and dependable though I can’t explain why…maybe I just need more time with Codex? Maybe because Codex UX is messier, lots of boxes within the text, artifacts are shown between the models reply and my text box.
10/ Codex or Claude, whatever you use…it is an incredible time to be alive. I had never written code prior to 2025.
11/ Humans are moving from makers to crafters. This is the OEM equivalent of software.
12/ The current UI/UX of AI is still in its DOS era (linear prompt <> response sequence)…will share my work on the next era soon.
For now : Claude builds, Codex reviews, I don’t sleep.
Cc: @bcherny @thsottiaux