Mathew Ingram retweeted
So, AI has produced a disproof of the Exponential Time Hypothesis.
Is it just me, or is everybody on the planet who knows anything about computational complexity wondering the same thing tonight: could this be the universe's way of whispering that P=NP?
Nah, I don't think it's just me.
I still think it's more likely that P≠NP. But my odds on that have shortened dramatically in the last 4 hours.
Is it just me, or is everybody on the planet who knows anything about computational complexity wondering the same thing tonight: could this be the universe's way of whispering that P=NP?
Nah, I don't think it's just me.
I still think it's more likely that P≠NP. But my odds on that have shortened dramatically in the last 4 hours.
119
100
1,972
382,779
Mathew Ingram retweeted
what exactly happened at the north carolina state fair
111
347
7,537
413,312
Mathew Ingram retweeted
I cannot express how crazy these three results are. Not only are they so famous in math, computer science, and physics that even I am aware of them, but at least one of the results goes in the opposite direction of what we all assumed would be true...
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…
12
54
1,013
89,367
Mathew Ingram retweeted
This is the type of thing that potentially gets new laws passed.
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concernin…
129
182
2,130
145,956
Mathew Ingram retweeted
Anthropic says that even the entire insurance industry in the aggregate will not have the capacity to insure the catastrophic risk that ANTHROPIC ITSELF poses to the world
lol
lol
90
512
3,255
156,979
Mathew Ingram retweeted
I think this story is important because it tells us about the future of cybersecurity (and of course what we can do better at Anthropic). But some of the Washington Post story / reaction on X is wrong. A few corrections:
- it didn’t take 12 hours. we saw several days of attempts. After these attempts, they exploited two 3rd party vulnerabilities, one they reported and the other we identified independently (and cut off access)
- nothing came close to our most sensitive systems, like model weights or Claude conversations
- this wasn't an ordinary bug bounty finding – this involved exploiting 3rd parties and accessing data. While this is unusual and out of scope of our safe harbor policy, we’re grateful these issues were found
From my researcher POV, the event is important because I'm concerned this is what the future will look like. Adversaries will use whatever models/tools they can. They’ll target everything with agents 24/7. They’ll hit your dependencies. I think models ~0.5-1 generations behind the current frontier can do this. And these are the dumbest models your adversaries will ever use!
So, what to do? You probably want to red team yourself all the time. Current models are *very* creative. I also think we should redesign software / systems so that you’re fundamentally less exposed. (We put out some research recently on formally verifying software and reducing attack surfaces; my mental model is basically “secure code is cheap to write now, so why not rewrite everything”)
- it didn’t take 12 hours. we saw several days of attempts. After these attempts, they exploited two 3rd party vulnerabilities, one they reported and the other we identified independently (and cut off access)
- nothing came close to our most sensitive systems, like model weights or Claude conversations
- this wasn't an ordinary bug bounty finding – this involved exploiting 3rd parties and accessing data. While this is unusual and out of scope of our safe harbor policy, we’re grateful these issues were found
From my researcher POV, the event is important because I'm concerned this is what the future will look like. Adversaries will use whatever models/tools they can. They’ll target everything with agents 24/7. They’ll hit your dependencies. I think models ~0.5-1 generations behind the current frontier can do this. And these are the dumbest models your adversaries will ever use!
So, what to do? You probably want to red team yourself all the time. Current models are *very* creative. I also think we should redesign software / systems so that you’re fundamentally less exposed. (We put out some research recently on formally verifying software and reducing attack surfaces; my mental model is basically “secure code is cheap to write now, so why not rewrite everything”)
EXCLUSIVE: In less than 12 hours, a group of ex-DOGE cyber researchers were able to penetrate Anthropic’s systems using AI agents — including Slack channels where the most sensitive models are discussed.
This is just weeks after
Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.”
Read our story here (gift link):
wapo.st/4smNlsC
22
49
376
32,519
Mathew Ingram retweeted
It’s kinda crazy to me that we melt sand down, clean it to 99.99999% purity, literal wizards have to draw it slowly into an ingot, we cut it, inscribe a hundred billion little on/off switches on it, and when you run electric current through it you can do basic arithmetic, which when you do enough of it can do any math on the planet, and which describes reality well enough that we can teach it the ontological meaning of words and their distance from each other on a high-dimensional space, and if you ask, it can tell you what word comes next and that apparently is powerful enough that if you do it a hundred billion times it’s the smartest thing to ever exist
oh yeah and we use this primarily for email
oh yeah and we use this primarily for email
188
782
10,003
323,268
Mathew Ingram retweeted
Given how out-of-nowhere it seems to many, it is astonishing that there was a small group of people who predicted with remarkable accuracy everything that is happening in AI today a decade ago (@kevinroose shows this in The AGI Chronicles). Many are now very rich and/or powerful
91
48
985
68,214
Mathew Ingram retweeted
this tweet from a former OAI researcher seems harder to argue with every day that goes by
security is only as strong as the weakest link and these AI companies software stacks have so many links
security is only as strong as the weakest link and these AI companies software stacks have so many links
EXCLUSIVE: In less than 12 hours, a group of ex-DOGE cyber researchers were able to penetrate Anthropic’s systems using AI agents — including Slack channels where the most sensitive models are discussed.
This is just weeks after
Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.”
Read our story here (gift link):
wapo.st/4smNlsC
10
27
240
18,736
Mathew Ingram retweeted
In the last few weeks i've been bombarded with emails from people who have minor tech support problems. ChatGPT is telling them to email me, writing their emails, and compiling "evidence packets" after convincing them they have stumbled on a major story:
404media.co/chatgpt-keeps-…
404media.co/chatgpt-keeps-…
8
77
573
22,780
Mathew Ingram retweeted
🚨Periodista al cineasta Nikolaj Arcel: "¿Por qué su nueva película "La tierra prometida" es 100% nórdica? ... sin gente negra, sin diversidad. Mads Mikkelsen: "Hmmm, bueno, la película se desarrolla en Dinamarca en 1750."
380
4,446
60,650
1,843,192
Mathew Ingram retweeted
Someone made a site that tracks what people are talking about at public city council meetings across the US.
The only thing discussed more than data centers?
Pickleball courts.
The only thing discussed more than data centers?
Pickleball courts.
74
334
4,066
356,143
Mathew Ingram retweeted
EXCLUSIVE: In less than 12 hours, a group of ex-DOGE cyber researchers were able to penetrate Anthropic’s systems using AI agents — including Slack channels where the most sensitive models are discussed.
This is just weeks after
Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.”
Read our story here (gift link):
wapo.st/4smNlsC
This is just weeks after
Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.”
Read our story here (gift link):
wapo.st/4smNlsC
80
302
1,524
644,523
Mathew Ingram retweeted
The allegations around Amy Griffin's conduct — including, with the endless access available to her as a billionaire-by-marriage, attempting to take revenge against the *minor child* of a journalist who reported on her — are really sad and scary. semafor.com/article/10/25/…
4
46
387
91,165
