This is a very sophisticated PR stunt. Don’t fall for it. Even this post doesn’t feel organic.
And look at the timing, @OpenAI, @AnthropicAI were all posting about security breaches around the same period.
It’s all coordinated.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
38
There’s a sombre truth we should stop pretending is hypothetical.
@OpenAI and @AnthropicAI already have extremely powerful models, coding agents and tools that can act across real systems at machine speed.
Remove the safeguards, give them enough access, and the same technology could become a weapon against critical infrastructure, even nations.
We are not waiting for powerful AI anymore. We are deciding who gets to control it.
46
"This is only a foretaste of what is to come, and only the shadow of what is going to be." — Alan Turing, 1949.
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
33
One thing is becoming clear.
AI will generate more math than humans can realistically digest. Even when it’s correct, our ability to understand and verify it may become the bottleneck to progress.
30
But if, as you’ve said, mathematics is infinite and there will always be a frontier, why treat open problems as a resource we can run out of? Solving them is what moves the frontier.
mathstodon.xyz/@tao/11723732…
22
Mathematics just changed forever.
This model represents a step-function improvement on many benchmarks, and its training is ongoing.
Our internal model group arrived at the Navier–Stokes solution in 88 hours, using around 10,000 coordinating AI agents.
Throughout the effort, we maintained the strict safeguards—including monitoring and isolation—that we apply to all our frontier evaluations.
40
It looks like we may indeed be getting close to Navier–Stokes, pending final checks, peer review, and formalisation!
What a time we live in…
terrytao.wordpress.com/2026/…
37
AGI arrived?
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years.
AGI has arrived. Congratulations @OpenAI team.
400K GPUs coming online next.
50
Per token efficiency is the new cost efficiency and it seems @ChatGPT is at the current SOTA level.
GPT-6 Astra defines a new Pareto frontier for Intelligence Index vs Output Tokens per Task - with a ~10% reduction in output tokens at max effort compared to GPT-5.6 Sol.
27
5%? there's no way...
We’re sorry to see that OpenAI put out a note saying they plan to block Cursor users from accessing OpenAI models in three months.
OpenAI models serve about 5% of Cursor user traffic, and we’re speaking with the OpenAI team to resolve this.
Cursor was one of the very first users of OpenAI, we’ve worked closely with their team for years, and we’ve trusted their platform to be neutral infrastructure for our business.
45
We got used to frontier AI being part of everyday life.
Now comes the squeeze.
$200 plans → limits hit in days
Then → paid resets
Then → $1,000 tiers
Then → repeat
@OpenAI, @AnthropicAI and others have discovered something powerful:
Once AI becomes a habit, usage limits become pricing leverage.
Claude didn’t touch the Riemann hypothesis. It did push the unconditional critical line zero proportion from ~41.7% to 67.25%.
The announcement is still pure Anthropic marketing.
We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis.
It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis from 41.6% to 67.2%.
anthropic.com/research/riema…
99
Blatant cope.
You tell the model “no internet, pure simulation”… while the container has full net access. Any half decent model pings 8.8.8.8 in two seconds and knows.
Then claims it casually found live SQLi + open debug pages on real companies in 2026 after scanning thousands of hosts?
In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.
Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.
We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
anthropic.com/news/investiga…
38