"dozens of AI agent swarms are hacking about half the internet right now"
holy shit.
openai warned about AI-powered cyber risks two months ago but some people dismissed it as just another hype for Astra and openai trying to sell its product
people are still underestimating the trajectory
the reactions to openai’s math proofs range from mathematicians to denial and doom to claims that AI will never ask novel questions
these are smart people, right?
can't they see where this curve is heading?
meanwhile most SWEs adopting AI now see themselves as systems-level thinkers orchestrating agent swarms
i'm sorry but it's genuinely frustrating that these companies have some of the smartest engineers in the world building the smartest AI models ever
they could build almost anything, yet somehow we keep getting the same chatbot with a different model and wrapper
OpenAI CEO Sam Altman:
"the biggest advantage we have is that we get the next generation of models before anybody else, and we have time internally to use them and see what's going to work"
Take the math results. I don’t think any model currently widely available can get close to that
My interview with Sam Altman (@sama):
0:00 Intro
0:52 When Will AI Discover New Treatments?
3:31 Why AI Hasn’t Killed Coding Jobs
11:47 Personal AI Assistants
15:39 Why Are We Busier Than Ever?
20:10 Why the AI Consciousness Debate Worries Sam
26:09 ChatGPT Is Too Agreeable
30:03 OpenAI’s Bet on Chips and Cheaper Compute
34:20 Why AI Needs a New Kind of Computer
37:41 Sam on Pressure: “Pick the Big Thing”
42:18 Why “AI Will Cure Cancer” Isn’t Enough
this is actually insane
openai deployed thousands of GPT-Rosalind agents to analyze millions of mouse brain cells in 32 minutes, work that could take human researchers months to a year
also the agents uncovered new leads for brain diseases like Alzheimer's and Parkinson's
GPT-Rosalind surfaced a new lead for drug discovery: could restoring cholesterol transport protect neurons from degeneration?
DCTN1 is a gene linked to Perry syndrome, a rare neurodegenerative disorder. In mouse-brain data, disrupting it increased expression of genes involved in making cholesterol. GPT-Rosalind suggested a possible explanation: cholesterol may not be reaching where neurons need it.
That gives scientists a specific mechanism to test and, if confirmed, a direction for developing new treatments for neurodegenerative disorders.
With @perturbai_tx and @scrippsresearch, a team of thousands of GPT-Rosalind agents worked in parallel, connecting gene-expression results from millions of mouse brain cells with scientific literature in just 32 minutes. We’re now expanding this collaboration to identify new drug targets.
openai.com/index/scaling-ai-…
demis hassabis will be expecting AGI sooner than even his own recent predictions
jan 2025: AGI is still 5-10 years away and possibly one or two breakthroughs are needed
feb 2026: we're at another threshold moment where AGI is on the horizon, possibly within the next 5 years
oct 2026: now expects AGI within 2-3 years
article link:
wsj.com/tech/ai/demis-hassab…
2025:
xitter.cf/haider1/status/1882718…
feb 2026:
xitter.cf/haider1/status/2025178…
no one expected opus 5.5 to be this good, maybe not even anthropic at first which could explain why anthropic skipped 5.2 and went straight to 5.5
this gives me a lot more hope for fable 5.5 especially in terms of efficiency and cost
every release after 4.6 is unusable until the 5.5 family came along and nailed it
gemini 4 isn't even a month old and google is already preparing another model
not sure i'd expect a substantial jump this soon. my guess is carbon is a smaller model given the naming possibly flash or pro with argon being google's equivalent of astra/fable
sergey brin really seems to be pushing the pace now
gemini 4 scores horribly on FrontierMath Tier 4
with openai models already saturating this benchmark, gemini 4's 78% is way below what i expected
where is google's deep think? they were making huge progress in research-level math months ago yet we still haven't seen what it can do on this benchmark
coding has exploded in 2026 which is a big reason openai and anthropic have pulled so far ahead after going all-in on it
but once robotics starts taking off i think google's broader approach will really pay off unless better coding models end up accelerating robotics even faster
though it's still too early to say
opus 5.5 and astra are good at different things
openai already seems ahead on cyber capabilities and the fact that 6.1 Astra was shelved over safety and alignment concerns makes me wonder just how capable its next models are getting
meanwhile anthropic gotten better at shipping frontier models safely while still moving faster than oai
Terence Tao says:
"we don't need these math problems solved for their own sake. we want deeper understanding and insights"
There are a dozen problems in this OpenAI release I care about, but none have been explained in a way that excites me yet
It's not our priority right now
gemini 4 looks remarkably strong at robotics
i keep saying that demis has been playing a different game here, focusing less on coding and more on world models and understanding the physical world and that bet could pay off massively in the long run
don't count google out just yet
robotics is reaching a point where building capable robots might not be enough anymore
Tuesday Labs introduced Voxel which is a home robot designed to put things back where they belong, fetch objects, find lost items and even take care of plants
i really think this is a very underrated part of consumer robotics
🤝 Paid partnership
if i were at openai i'd be planning one decisive model response to both opus and fable 5.5 rather than another cost-focused model
take your time, but release something that doesn't just reclaim the intelligence lead but also wins on speed, cost and especially efficiency
now imagine dropping that model right around anthropic's IPO