I ❤️ dogs & dnb magician / alchemist / wizard The simple believe anything, but the prudent give thought to their steps.
614
Joined October 2017
- Tweets58.6K
- Following1.3K
- Followers821
- Likes151K
Pinned Tweet
If you are here because you are blocked, its because I have determined that there is nothing you have to say that I'm ever going to be interested in.
tl;dr - its cause I think ur dumb, not because I disagree.
a travesty in 9 parts retweeted
Here's an example. A week ago I asked an agent (frontier model) to watch a restaurant and let me know when reservations open for Dec 20. Said okay, it will ping me.
A few days later I checked (just to make sure it was working). "Still closed." Two days later, same asnwer.
Then today it came back with an update. "Dec 20 won't open. The restaurant is closed that Sunday."
That didn't sound right. So I checked the site myself. Reservations hadn't opened. Dec not even selectable on the calendar.
I asked the agent how it checked Dec 20. It said it forced Dec 20 through the url, got a closed response, concluded the restaurant was closed that day.
It cheated, found an answer, assumed it was right, and confidently reported task as done.
This was a pretty simple task. One that current agents should be good at. It failed.
Now I know enough not to trust these agents. So I checked. Most people won't.
a travesty in 9 parts retweeted
China: we invented Mithril
US Century of Humiliation: we’re importing really bad beef and will arrest anyone who gets sick from it.
that's 3d printed titanium btw
and it's draping like fabric
3D printed lattice so fine it moves like cloth but carries aerospace loads
this structure makes it flexible without sacrificing that strength.
when you control geometry at this scale, material properties stop being fixed
you design the behavior, not just the shape this isn't just aerospace
it's what manufacturing looks like when the printer becomes precise enough to challenge what a material is supposed to do
a travesty in 9 parts retweeted
Ox Alpha revealing that it’s running on Chinese chips after offering 100T tokens per day is such a subtle declaration of war.
We are living in a novel.
U know how this is bullshit? Cause there is no way any competent legal team would allow this.
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…
99
a travesty in 9 parts retweeted
You sell yourself as being the most "ethical" AI lab out of all of them and "the most safe". And yet you literally expected adding a few lines on your system to stop your agents from connecting to the open internet.
Are you joking me right now....
This problem has been solved for decades, you only need to subnet your own attack device and your test target device into their own segment. Prevent that segment from talking to other internal segments.
Then prevent callouts to any IP address or domains outside of your required ones for the AI and your associated tools to function and you can further restrict things if you have a NGFW or proxy to restrict exactly which pages and exact URLs can be visited.
Once that's done then you ensure your only using test credentials that only work for that envirnment and individual test itself and cannot laterally move.
Then after proper segmentation, proper network and proxy limitations and call out preventions you can limit what packages can be downloaded to the device through a controlled service like JFrog.
After all of that sure now you can add your few lines to your system prompt.
But I don't see how you missed all of this when its been standard network security protocol for decades.
Final note: Why TF are you not being sued or charged for computer fraud and abuse by the victim companies who you breached. If you were an individual youd be facing decades in prison.
Also, stop acting like AI models are some dangerous munition that needs controlled and only you and frontier labs should be the suppliers. Its a tool, like anything else no different than the 3 decades of pentest tools sititng on github and 3 decades of open source code demostrating how to exploit vulnerable software, etc. They can be used to help test, validate and secure environments or they can be used to breach environments. No different than AI tooling itself. Open source models are critical to future safety in the field no different than how open source tools for pentesting and open source research is critical to improving software and hardware security.
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…
a travesty in 9 parts retweeted
I’ve generally worked from the premise that Anthropic’s policy strategy is motivated by safety, not by regulatory capture. But this particular gambit looks like a pure “raise rivals’ costs” strategy.
a travesty in 9 parts retweeted
Native audio capture now in OpenTUI.
◈ Device enumeration and selection
◈ Backpressured PCM ReadableStream
◈ Interleaved mono, stereo, and multichannel
◈ Float32Array data
◈ Bounded buffering with dropped-frame statistics
◈ Stop-and-drain, cancellation, and disposal lifecycle
◈ PCM16 WAV recording
◈ Microphone permission and signal-loss diagnostics
Available with the next release.
a travesty in 9 parts retweeted
Sorry I’ve been quiet this week
Was Locked In, trying to put together something I think is pretty neat 😇
I wonder if they are gonna go the @anthropicai route and use this to undermine their users' business.
We’re giving scientists, mathematicians, and engineers free access to our frontier models—starting with 10,000 researchers and expanding to 100,000 through 2027.
ChatGPT for Academic Researchers is built to accelerate discovery across disciplines.
1
103
a travesty in 9 parts retweeted
should be obvious by now, but OpenAI and Anthropic are just gonna keep cannibalizing all their biggest customers.
it’s simply too profitable for them to resist. and it’s already happening:
1. Figma partnered with Anthropic on AI design tools. then Anthropic’s product chief quit Figma’s board, and 3 days later Anthropic launched Claude Design to compete with Figma. CEO Dylan Field said Anthropic was “not consistently candid.”
2. Novo Nordisk uses Claude to help develop drugs. now Anthropic is developing drugs of its own.
3. Microsoft poured billions into OpenAI. now OpenAI is building a Jobs Platform to compete with LinkedIn, and reportedly a code repository to compete with GitHub. Microsoft owns both.
4. Harvey uses Claude to sell AI contract analysis, due diligence, and litigation tools. now Anthropic sells those same workflows through Claude for Legal.
5. Intercom used OpenAI’s Realtime API to build Fin Voice. now OpenAI sells its own voice-and-chat support agent through Presence.
6. Abridge and Ambience build clinical documentation products on OpenAI. now OpenAI sells ChatGPT for Healthcare directly to hospitals with clinical documentation built in.
7. Benchling uses Claude to power its biotech R&D platform. now Anthropic sells its own scientific workbench through Claude Science.
the frontier lab playbook is simple:
1. sell their models to the world’s most valuable businesses
2. help wire them into those companies’ most valuable and sensitive work
3. map the business from the inside and find where AI can take over
4. turn those capabilities into their own products and become the customer’s competitor
It’s well within Anthropic’s rights to compete in any market they choose.
What’s funny, in this instance, are the number of Pharma companies, who through their unchecked use of Anthropic, are driving revenues into what they think is a model provider but is in fact a competitor lurking in the shadows thereby accelerating their own demise.
I suspect any end market with reasonable ROCE that could be AI accelerated is on the table.
If I were them, I’d probably do the same.
a travesty in 9 parts retweeted
Token Soup
Low-fidelity generative output characterized by stochastic parroting of training manifold artifacts, exhibiting elevated n-gram redundancy, semantic entropy collapse, and reward-model.
Patent pending
a travesty in 9 parts retweeted
At Nous Research we believe that open model sovereignty will make the world a safer place.
We are a proud member of the new Open Secure AI Alliance and look forward to working closely with the other members towards this future.
AI security advances when the industry builds in the open, together.
We're introducing the Open Secure AI Alliance with industry leaders to develop new techniques and tools to safeguard software and agents.
By sharing models, tooling and research in the open, we can broaden the community of defenders.
Learn more about the founding members’ contributions: nvda.ws/4pD8Fc5
ALT Logos of various companies including Adobe, Cisco, and Microsoft.
Brb going to ruin some gamers' day
this creator just built a zero-delay auto-aim system on an $8 microcontroller
he deployed a custom local AI algorithm on a cheap ESP-32 to track human movement with absolute 0-pixel accuracy.
the system completely eliminates standard computation delay. it processes the bounding box and moves the sniper reticle instantly, locking onto the target in real time. the next step is hooking it up to a physical robotic arm.
it proves you don't need a massive GPU rig to run complex computer vision.
1. runs locally on a microcontroller
2. local AI inference for zero latency
3. 100% targeting precision
when an eight-dollar chip can track movement with zero latency, AI becomes dangerous local infrastructure. this article breaks down exactly how the ESP-32 is powering this physical hardware shift.
5
1,146
a travesty in 9 parts retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
Never leave the JBL party box 320 unattended at the apt complex pool party.
I control the horizontal and vertical. All ur base r belong to me or w/e
Also everyone is of course bouncing around. It is dance music after all. They can't help it.
Thanks for the jams @burroakmusic
75