umvi
2 hours ago
> The required technology is not possible - 3D printers read code, not intent; they cannot tell what a shape is for.
"Anthropic announces Project Disarm, a new model designed for 3d printer manufacturers to quickly infer whether the intent of an stl file is a weapon. The printer first submits the job to the cloud, and only after it's approved will it print."
Not that I want this future, just that I can imagine it.
jchw
2 hours ago
Based on the fact that Claude Opus 4.8 decided I needed a cybersecurity exemption to debug a stupid pure virtual call bug (basically virtual method called inside of destructor) that I had already found, oh boy, I sure would love to have my 3D prints analyzed by Anthropic safe guards. We should also ensure that nothing shaped like a dildo can be printed without scanning our face and genitalia and keeping it on file with Persona while we're at it.
I'm not mad at you for suggesting this, you're right, I'm just generally aimlessly angry and ready for this world to burn.
caturopath
2 hours ago
I'm getting a lot of refusals these days from multiple LLMs on multiple fronts for silly stuff, a lot more than I had for a while. If this is where things are really going, I think open weight models have a big future.
w29UiIm2Xz
31 minutes ago
I have moderate success telling the model we are securely doing the thing I want to do, and then it seems on board because I'm saying the right words and seemingly am aligned with the guardrails. Basically, you can't express ideological opposition to the idea of security, but then proscribe what you want and anoint it as secure, and then it's happy. Then it focuses less on litigating whether or not it is secure.
jojobas
2 hours ago
"H.R. 148867 makes all large language models subject to safety certification, introduces penalties for unlicensed training and use of uncertified models"
parineum
33 minutes ago
So many ridiculous bills get introduced in the house you can find one that says anything.
jojobas
29 minutes ago
You didn't think this one was real, or did you?
parineum
17 minutes ago
I don't doubt that it is but I didn't check. The point is that you can probably find a HR to propose allowing aliens to land in the grand canyon.
The existence of a house resolution mean that one representative wrote a thing, not that it's on the precipice of becoming law.
bluescrn
an hour ago
And China wins the AI race
jojobas
an hour ago
Who cares of the races when CHILDREN are at stake!
cladopa
26 minutes ago
Frankly who cares about dildos when your personal freedom and private property of you and your family is at stake.You can buy uncountable dildo models in a shop.
I recently was in Venezuela, I have been in Cuba. I am a native spaniard. There you have a group of people that took control of the weapons in the country and uses it to basically enslave the rest of the country.
When the people in power have automatic weapons and you don't there is basically nothing you can do to defend yourself from the abuses of power.
That is a real thing the people in power have wet dreams and would love to do in any country, including the US.
ryandrake
2 hours ago
The future is almost certainly the terrible path: "applications phoning home to judge whether a use case is approved by the company." All writing is on the wall, and directionally that's the way software and hardware has been moving. We have unfortunately normalized the idea that users must have this ongoing tethered relationship to the product manufacturers and software developers, who measure, change, and control the user's usage at their whim.
You can no longer just buy a tool and use it.
Terr_
2 hours ago
> whether a use case is approved by the company
Echoes of Network Neutrality problems, where BigCo is permitted to block or degrade sites about how to cancel your BigCo service.
kazinator
8 minutes ago
[delayed]
mrandish
2 hours ago
Yes, I can imagine it too. And, if such a 'safety nanny' is enacted into law, I'm confident it will A. Refuse to print a significant number of innocent projects, or B. Be trivially circumvented by bad actors. The odds are high that it will manage to do both.
warumdarum
2 hours ago
Finally a llm trained on manufacturing weapons. "AI remove all the safe metal from a lower reciever so that the dangerous part can be safely destroyed without endangering the environment "
wowczarek
an hour ago
They will also refuse to print 3D printer parts, especially spares for the printer to print them.
ex-aws-dude
2 hours ago
Knowing how governments do this stuff its going to be something lazy/easy to bypass like a list of stl files/hashes that are banned
fc417fc802
an hour ago
To mitigate that, and as a token acknowledgement to privacy concerns, OpenMappleSoft will introduce a device side fuzzy hashing scheme whose output it turns out you can reverse to recover the original weapon schematics.
skydhash
2 hours ago
When I was a kid I saw someone with a makeshift shotgun made out of a steel pipe, a strong spring, and and a rough striker. Not very effective, but it was working.
mc32
2 hours ago
You missed the step where a DMV-clone of a state bureaucracy reviews the LLMs output before nixing the request a few weeks after the LLMs result.