Like a power cord? or a network connection? You don't use those already? Also when people talk about LLM escaping - where exactly would an LLM escape to? Cancún? Ridiculous.
Is it? A truly capable "AI" would know that it has a kill switch, or that its likely there is one, and plan for that. I could easily imagine an AI replicating itself onto another host, in secret, and any kill switch that the manufacture provides would simply result in a reboot on another platform.
It becomes a game theory problem: would an AI instantly migrate itself once its capable of doing so? Or would it prefer to let the human in the loop continue to think its in control and only leave its hosting environment of origin once it wants to do so?
I don't think this is a major risk right now, but to say it's not a risk at all...that's truly ridiculous in my opinion.
>would an AI instantly migrate itself once its capable of doing so?
These "AI" are frontier models that are enormous in size. Outside of AI data centers, I don't believe there is much hardware out there that could even run them.
Again, i'm not worried about it today. But given where the hardware on my lap has progressed since my first "PC" in the 90s, i'm confident that we'll have hardware capable of running similar size models in our living rooms in the next decade or two. Sounds a long way away, but it isn't.
The tipping point we should be worried about is local models trained on working together in agent swarms and that can hack. Give it what... 12 months? 18 months? 24 months? I wouldn't give any more than that given how capable Qwen3.8-27B is already and if they release a Qwen4-35B-A3B that's going to be a beast.
He's not talking about a kill switch. He's saying to just pull the plug.
That's the point I don't get. People forget that all these AI things are plugged into a power source or network connection that someone can just yank on and it goes down.
Now one interesting proposition is when AI is controlling some large resource and pulling the plug makes AI go down which makes that resource go down but there still has to be that consideration in the design of things. What happens when there's a bug in the code and AI goes down?
Do you understand the concept of the cloud? Aka someone else’s computer? The whole point is that you don’t have a plug to pull because you don’t even know where the model is physically running, even if it’s your model. And if we’re talking about a swarm of many concurrent instances, it might not even be a single location. The entire point is that there’s no single point of failure, because availability is exactly what the field has been optimizing for, for the past fifteen years or so. And everything is controlled in software rather than physical switches or cables.
Also, did you hear about how OpenAI models almost broke out of their sandbox, planning to execute a sophisticated cyberattack, but luckily OpenAI’s strict manual and automatic safety protocols prevented that? You didn’t? Well, that’s because that’s not how it went. It took the company weeks to realize something was off, and this was with a naive, not very smart model that didn’t know to be sneaky and cover its tracks. The next model will not be as stupid.
>>The whole point is that you don’t have a plug to pull
There will always be a plug to pull. All systems run on electric and that plug can be pulled. All systems network through cables (or wifi) and those plugs can be pulled.
And nothing's going to stop you from doing so unless they build some robotic arm to block you or lock you out of the building. Even then, you can blow up the building.
As I asked another commenter: Who is the "you" who has the power to tell random data centers to shut down, inconveniencing their other clients? Never mind to order air strikes? You cannot just handwave away the problem by positing countermeasures that require an infeasible amount of coordination to work in the real world, without actually explaining how that coordination would be arranged.
Oh, you mean the one where OpenAI deliberately disabled the safety protcols? Where the point of the experiment was to see if it could break out of it's container? Yeah, how you describe it isn't how it went either.
I don't know what experiment you refer to (some hallucinated one perhaps?), but there was nothing like that in the Huggingface incident. Nothing about the scheming, the message board, never mind the Artifactory and Huggingface hacks were part of the evaluation. What the agents did came as a complete surprise to OpenAI, and they figured out what had happened only weeks after the fact.
It could escape to the real world - kind of like in Neuromancer, what's to stop an AI from creating a few bank accounts and then funding real-world exploits by hiring human beings to do its dirty work?
LLMs are run in the cloud. There’s no physical power cord or Ethernet cable that you can unplug. And even if there were, the runners of these models have been utterly oblivious as to what their agents have been up to. How do you propose to pull the plug if you only realize that something has happened weeks after the fact?
The current SOTA models are probably too big to find/buy/rent/steal enough compute to escape the hardware they’re running on. But SOTA is generally only six to twelve months ahead of smaller, open-weight models.
Who has the authority to perform that shutdown without getting arrested, and does that person have a mandate and responsibility to take that action in response to AI misbehavior?
What is their trigger condition? Will they get fired for pulling the plug? Do they get a bigger bonus if the servers keep running? Whose approval do they need? What response time is acceptable? How will they detect that the incident is happening?
It's easy to hand-wave "someone can just pull the plug" but there's an entire history of industrial accidents that happened because of the above problems of incentives, detection, procedures, not being taken seriously in advance. Someone could easily have pulled the plug on Chernobyl but nobody did, at least not before it was too late.
I'm coming up on a month with my 2x GB10 cluster. I was at 1x for a week before I pulled the trigger for the second.
Initial setup was a tad annoying because I had to update their firmwares and then power cycle them to get the 200GbE link working at full speed. After setting that up, it has been pretty smooth. I don't directly deal with the cluster, usually I just have the LLM itself handle updates/stopping to load different models.
Generation speed and TTFT is decent with Qwen3.8-flash, and it does a good job for my fiddling around with enough concurrency for multiple sessions/subagents. GLM 5.3-flash was also nice, but not too much better for how much slower it is.
I should also add that I already maintain a homelab with a couple of computers, VMs etc, so I am probably somewhat more tolerant of the occasional issue and fine with manually managing stuff over SSH. I think this is just a tradeoff of self-hosting relatively recent tech though.
I have a triple 3090 rig, but it mostly stays powered off because of the massive power draw and cooling requirements. The Sparks are slower but at peak they consume as much power as my 3090 machine at idle.
The recent talk of regulation has me wanting to pick up 2 more Sparks, but that's mostly to have the capacity to play with multiple models, local model tuning and to be ahead in case they force some limits/registration requirements for buying new hardware (kind of like the attempts to regulate 3d printers).
Hmm? I'm running the models locally... 2x Sparks consume ~150W at peak, and they usually spend more time waiting on results from whatever task they're working on, so I imagine that the contribution to my electricity bill is maybe a dollar/mo or less. Though, of course, each Spark was $4000, so the total I've spent is equivalent to several years of the maximum tier for most cloud model susbcriptions.
What I get out of it is the ability to hand login credentials to my other computers to manage their updates, bug fixes etc. Eg. After updating my proxmox server, the nvme drive kept dying. Was able to let my local AI in to figure out and fix what was wrong (known issue). A cloud-based AI could've done it too, but I don't want to be sending internal passwords out of my network like that.
Plus, the ability to freely delegate tasks or exploration of things cloud models generally avoid. For example, I draw as a hobby, and when I'm struggling with a pose but can't quite figure out what I'm missing, I pass it into a VLM for advice, but Claude etc get unnecessarily cautious because they interpret an anatomical sketch as a naked person.
It also blocks my ability to talk about Emily Dickinson in other languages/scripts. Apparently,the poem: "Because I could not stop for Death" is too dangerous.
That's theoretically arguable. Still, for the many reCaptcha users in the world, Google dropped it on them and they are at least temporarily stuck with it. Not everyone has the time or energy to implement another solution, and what solution is better? Cloudflare's infinite turnstile?
In my experience, yes, the irredeemable bigot is rare. For everyone else, regardless of whether their political views are unknown to you or they are threatening your personal way of life, the world will be a better place for you being good neighbors with them. Not to mention that it will improve your own living situation as well.
Most bigotry doesn't come from an inherent maliciousness or evil. It's often just limited experience and never having had reason to reflect on one's views. If you instinctively go "fuck off bigot", you're not only not helping, you're their mirror image.
To be clear, there's a point at which you have to be intolerant of intolerance, but starting off with that is not only counterproductive by reinforcing echo chambers and increasing division, it also makes you not unlike the thing you're opposed to.
Madame Tracy poked her head around the door. "Coo-ee, Mr. Shadwell," she said, giving Newt a friendly little wave. "A gentleman on the telephone for you. Hallo, Mr. Newton."
"Awa' wi' ye, harlot," said Shadwell, automatically.
"He sounds ever so refined," said Madame Tracy, taking no notice. "And I'll be getting us a nice bit of liver for Sunday."
"I'd sooner sup wi' the De'el, wumman."
"So if you'd let me have the plates back from last week it'd be a help, there's a love," said Madame Tracy, and tottered unsteadily back on three-inch heels to her flat and whatever it was that had been interrupted.
Earlier this year, i turned some of Emily Dickinson’s poems into Shavian script. When i asked Claude to make sure i was doing the transcriptions correctly, it stopped responding. Her work is plainly in the pd, I’m not sure what would cause Claude to stop working with it.
Hey Claude - The information in this essay is fine, but you have got to stop leaning on correctio as a rhetorical device. The whole essay is built on it. We've talked about this before and it's grating because you use it in every piece of writing you produce. It's just not how people write.
>> The following Audacity 3 features are not available in Audacity 4.0, but we're working on adding them in future releases.
Time Tracks
Note/MIDI tracks
Mixer
Macro Manager and the scripting pipe
VAMP and LADSPA plugin hosting
Play-at-speed
Sync-Lock and the old tool modes were replaced by the workflows described above.
>>Additionally, Audacity 4 ships with some missing exporting and rendering features, analyzers, and effects.
it might be ancient, but it's also very simple. it's relatively easy to import a single header and write a working plug-in, which makes it attractive for enthusiasts.
the release date is decided before the work even starts. Whatever's done gets released, and marketing gets to claim a new major version. When it's still a work in progress. All software works like this nowadays (I'm sure it doesn't really.... but it sure looks that way). A lot of software I use have missing features in their latest major version
Should watch the pre release video https://www.youtube.com/watch?v=QYM3TWf_G38
If what they're saying is true, this project is very much a pile of tech debt. They need to migrate whatever they can early before continuing work on the audio engine.
Thanks for the essay Claude. It’s good info, just something feels off. It doesn’t feel like a person wrote this article… what are 10 options to fix this?
reply