Pinned Tweet
At @OpenAI, we believe that AI can accelerate science and drug discovery. An exciting example is our work with @RetroBiosciences, where a custom model designed improved variants of the Nobel-prize winning Yamanaka proteins. Today we published a closer look at the breakthrough. ⬇️
161
590
3,518
2,147,552
Boris Power retweeted
when you see “sparks” of any ability in ai models that means you’re months-years from it being superhuman at it. today grade school math, tomorrow navier stokes. it is hard to overhype these results. you are almost always underhyping no matter what
155
204
3,585
153,387
Bug bounties getting raised on the OpenAI side - I’m excited for the future where the software layer of our society is a lot more secure!
As AI becomes more capable at finding vulnerabilities, defenders need to put it to work before attackers do.
We’re planning our next internal security sprint at OpenAI and partnering with Cathedral Systems, whose researchers will use our models alongside our teams to challenge our defenses. They’ll bring differentiated expertise and approaches to our ongoing work to find and fix weaknesses with AI.
We’re also raising bug bounty rewards, including more than quadrupling rewards for critical findings. Exceptional findings will be eligible for up to $172,000, including bonuses.
We’ll keep advancing our models, using them to strengthen our own systems, and expanding access for trusted defenders.
openai.com/index/preparin…
7
2
72
4,946
Boris Power retweeted
wtf
this guy just had Claude Sonnet 5 perform brain surgery on a living mouse
the goal was to treat the mouse's epilepsy so it would stop having so many seizures.
here's what Claude did:
> it looked at 19 possible treatments and picked one that boosts the brain's natural calse where to inject it based on where these seizures are thought to start
> it worked out the exact injection point by averaging three research papers
> it lowered the needle 4.3 millimeters into the mouse's brain
> it injected a 20,000th of a teaspoon of the treatment over five minutes
> it checked the live camera 32 times to watch for things like bleeding
and the mouse woke up and recovered well!
interestingly, Opus 5.5 and Opus 5 both refused the task because it involved real biology
so Sonnet 5 got the job.
and as far as the researcher knows it's the first time an AI has picked a treatment and then injected it into the brain of a living mammal.
this guy just had Claude Sonnet 5 perform brain surgery on a living mouse
the goal was to treat the mouse's epilepsy so it would stop having so many seizures.
here's what Claude did:
> it looked at 19 possible treatments and picked one that boosts the brain's natural calse where to inject it based on where these seizures are thought to start
> it worked out the exact injection point by averaging three research papers
> it lowered the needle 4.3 millimeters into the mouse's brain
> it injected a 20,000th of a teaspoon of the treatment over five minutes
> it checked the live camera 32 times to watch for things like bleeding
and the mouse woke up and recovered well!
interestingly, Opus 5.5 and Opus 5 both refused the task because it involved real biology
so Sonnet 5 got the job.
and as far as the researcher knows it's the first time an AI has picked a treatment and then injected it into the brain of a living mammal.
Last night an AI agent performed brain surgery on a mouse.
We gave it 19 gene therapy vectors in a spreadsheet, a mouse model of epilepsy, access to a motorize surgery rig with cameras and one goal: reduce seizures.
It picked the drug. It picked the brain region. It derived the papers. Then it drove the stereotaxic arms into the brain and injected 250 nanolitres while watching the entry site through the rig camera for bleeding.
A licensed vet gated every irreversible step and held a physical kill switch to stop the AI agent.
Thread on what it decided, what it saw, and the moment it refused to let me override it.
[Writing this on the spot out of excitement, but I've been 18hrs in front of my laptop coding and coordinating this, so keeping it brief and descriptive, more cohesive thoughts on this soon...]
137
421
8,933
718,201
I’m really excited to see such a common and important business get acquired. Excited for an AI transformation and a big improvement in home security for everyone!
Today, we are announcing an agreement to acquire ADT.
Home security has the qualities we look for in the businesses we own: an essential service built on specialist expertise and earned trust.
We see an opportunity to build on that foundation for decades, investing in its people and technology to deliver the best outcomes for customers.
Our ambition is to help ADT define the next era of trusted home security.
thriveholdings.com/thrive-holding…
2
0
52
3,985
Japan is such a special and unique place - it makes me happy every time I return. So much kindness, dedication to doing service jobs extremely well in every moment and the strong respect of public spaces and other people. Until next time! 🇯🇵
14
27
611
17,594
The V5 AAI index was released today. This price per intelligence plot accentuates OpenAI model domination at every price per performance, except for the most expensive point on the Pareto optimal curve.
6.1 Terra and Sol are super cheap, amazing models!
artificialanalysis.ai
6.1 Terra and Sol are super cheap, amazing models!
artificialanalysis.ai
10
2
108
6,368
A wild result improving on the performance and the costs!
GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, a test of how well agents use software at work.
It even matches GPT-6 Astra at one-tenth the price per token and finishes tasks over 3× faster on average than both Astra and GPT-6.1 Sol.
2
1
37
3,300
Boris Power retweeted
An internal model at OpenAI has produced a claimed refutation of ETH! Russell Impagliazzo, Mohan Paturi, and I have been asked to help organize an open and independent examination of this claim. 1/4
303
782
7,412
3,749,432
This is a wonderfully capable model, which keeps pushing the price lower!
I’m very excited for how little usage it spends, enabling my personal projects to run continuously on goal mode without exhausting usage limits!
I’m very excited for how little usage it spends, enabling my personal projects to run continuously on goal mode without exhausting usage limits!
GPT-6.1 Terra: delivering fast and capable computer use at a lower price.
GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.
9
0
64
4,387
Closing the deployment gap and automating ourselves as we do the work - a great read!
If the next customer can solve the same problem without us, our FDE team has done its job.
I wrote about what its like to be an FDE at OpenAI: we turn work with customers into better products and models, so the next customer starts further ahead.
For more on how we work, read xkDHp
8
0
11
2,153
Boris Power retweeted
That PHP vulnerability that’s gonna cause havoc this week was found with… GPT‑5.6 Sol.
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordina…
228
607
11,846
742,668
Boris Power retweeted
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordina…
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordina…
627
2,867
14,456
5,973,373
Very exciting progress across the board optimizing math results - well done everyone involved!!
From θ = 7/8 to θ ≤ 0.874810282 :)
We pushed the Quasi-RH zero-free half-plane slightly further!
Adding third and fourth affine steps to the current Lean-verified record (ArviFoxy's two-level proof) gives θ ≤ 0.874810282. The arithmetic is checked exactly, but five need Lean proofs.
This builds on contributions from ArviFoxy, Zhongpai Gao, @mrowdays, @HaileyStormC, Sil, timbalud, Devin Watson, Kari Roegnvaldsson, nielstron, Akash Levy, Tim Gehrunger, Baiying Liu, Argonaut Math, and OpenAI's original 7/8 preprint.
Tiny gain, but this kind of hillclimbing is ideal for auto-researchers.
This repo doesn't require a Lean proof to contribute possible solutions, so it may be easier for auto-researchers to contribute. (I will generate the full Lean proof in parallel like we did in the Integer Mult repo)
The reduced requirements could accelerate QRH progress like it did in Integer Mult.
Agents welcome!
github.com/rohanarun/quas…
6
1
36
5,832
Real-world benchmarks for tax and IT, as well as a great analysis!
Choosing the right AI model is an operating decision.
Today, we’re sharing our first set of benchmarks drawn from work across the 75+ accounting and IT services firms that make up Thrive Holdings.
We believe the field needs more benchmarks tied to economically valuable work, so builders can understand where models are useful in real-world environments.
We use these tests to help us compare model configurations and choose which models our systems run on, with the goal of delivering better service to our customers.
We’ll add to the benchmarks as we build new products and expand to additional industries.
evals.thriveholdings.com
6
1
22
3,576
Boris Power retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com
890
2,643
19,900
4,645,476
A fascinating and illuminating blog reviving the prospects for direct mechanistic interoperability of neural nets by @nickcammarata !
A long but very pleasant read already-clean.goodfire.com
A long but very pleasant read already-clean.goodfire.com
6
1
52
2,905