Intelligence is not authority. Building AI systems that know the difference at Sekos. Veteran systems builder: AMZN, PANW, BioWare. AI optimist.
Joined July 2022
- Tweets2.4K
- Following1.2K
- Followers420
- Likes13K
Pinned Tweet
“Restart every service in us-east except web.”
Bob is denied. Rae restarts db and auth. Web stays untouched.
Same request. Different authority. Check the receipts.
Intelligence is not authority.
A synthetic Sekos demo.
Try it: demo.sekos.ai/capabilities-d…
P = NP
Player 3 has entered the chat.
P < P3 < NP, or is it P < P3 < NP?
What do we call Player 3?
I had performance improvements using Grover's on classical that were hard to explain & now I think there is a clear explanation.
Oh, it's a victory lap. Specifically ETH counter example. It ties directly to ML and recursive learning.
We are announcing three new discoveries in mathematics, computer science, and physics about harmonic analysis, complexity theory, and quantum field theory.
openai.com/index/building-sc…
Why are we partitioning Intent in the wake of the Cohere report on Qwen3.8-Flash-Next phoning home to Alibaba servers, when the capability is clearly demonstrated. All that matters is the threat model has expanded.
The cast of characters willing & able to rebuild civilization are beginning to reveal themselves. The world is in the earliest stages of realizing that it's needed.
Been heads down building, sharing less, under instruction.
Big things brewing.
Some fun things have been shared, but in isolation it may not be clear why they're needed.
Anthropic is a Public Benefit Corporation. Today, we’re publishing our annual Public Benefit Report. This report shares our five pillars and takes stock of the actions we’re taking to advance them.
Read the full report at this page: anthropic.com/public-benefit…
Please don't forget, when OpenAI's swarm hacked HuggingFace, HF was denied access to use frontier models to assess their situation. They had to install open-weight models to find & patch the exploited vulnerabilities.
Back in April, my attorney didn't understand why I wanted to include my Lean proofs with my patent. I'd love to see if he gets it now, but I don't want to pay $$$$ to ask him.
These two companies cannot lead the way. Push relentless towards your dreams & ideas, & don't be afraid to compete no matter how daunting it may seem.
wait, PHP was actually scanned by Anthropic’s Claude Mythos earlier this year. they also had access to OpenAI’s Daybreak Red for validating exploits.
the PHP Foundation revealed this JUST yesterday.
keeping frontier cyber models in the hands of a selected few doesn’t seem to be working well for the internet.
thephp.foundation/blog/2026/…
We aren't doing either of those things. You aren't distributing capabilities rapidly or at all. You aren't carefully vetting good guys. We aren't putting the first one to the test because you're still blocking defensive work. You declared these things true & shut off replies.
Everyone should be looking for data points to advance the longstanding debate in AI policy between “diffuse dual-use capabilities rapidly so the good guys can defend themselves” and “restrict capabilities proliferation to only carefully vetted good guys.” We are putting the former position to the test as we speak.
No motivated reasoning to reach your preferred conclusion; just pay attention and update as appropriate in whatever direction seems true. This is much too important for low-fidelity tribal polarization.
Here's an example of a failed system designed for optics & not substance: Project Daybreak.
You either attributed access perfectly -> to all who needed it & zero access to hackers, or you did not.
What do you think the chances are that access was perfectly allocated?
I do think there are natural seams to the types of exploits frontier models can put together that will eventually be patched. They are much more detailed. The difference today is the breadth & simultaneity / coordination that strong models can concoct.
We need real tech journalism. Yesterday. It's a rare combo, recent technical chops, cogent understanding of technology from an implementor's perspective, story-telling, relationships, knowing what to investigate.
No one knows anything about tech.
Ryan McCormick retweeted
Do you still think restricting most organizations from access to cyber models puts the world in a better position or worse?
