Please or to access all these features

AIBU?

Share your dilemmas and get honest opinions from other Mumsnetters.

Am I being unreasonable to say we should be doing more to prevent AI wiping out humanity? YANBU - we need to do more YABU - it’s too hard to solve this problem so we just have to let AI kill us all/it’s never going to happen anyway

147 replies

Lukylukyluky · 11/09/2026 09:45

Is there a way to switch off all the electricity that AI runs on - if it goes rogue and starts to unleash nuclear weapons for example?
Are AI experts attempting to teach AI to behave in a moral and ethical way?
Are tech experts putting enough safeguards in place - to protect us from plausible dangers eg bad agent’s engineering deadly pathogens?
Is the government prepping for an AI induced breakdown on infrastructure on a massive scale?
Can you think of ways we can help ourselves?

OP posts:
Thread gallery
8
Krevlornswath · 11/09/2026 17:59

Lukylukyluky · 11/09/2026 09:45

Is there a way to switch off all the electricity that AI runs on - if it goes rogue and starts to unleash nuclear weapons for example?
Are AI experts attempting to teach AI to behave in a moral and ethical way?
Are tech experts putting enough safeguards in place - to protect us from plausible dangers eg bad agent’s engineering deadly pathogens?
Is the government prepping for an AI induced breakdown on infrastructure on a massive scale?
Can you think of ways we can help ourselves?

I work in AI safety and Human Data Ops research, so I can loosely answer most of these from my experience as somebody who is not privy to top level projects. Obviously, this is one person's perspective rather than a definitive view of the industry and will keep it broad in line with NDA.

1)Is there a way to switch off all the electricity that AI runs on - if it goes rogue and starts to unleash nuclear weapons for example?
A) "AI" isn't run from one central power source, so not in that sense. In real terms it would require the specific system to be taken offline, its credentials revoked, isolation from a network etc. There are many layers to the process. We do test for this eventuality and close-down is possible because human oversight is currently, retained. The AISI has also recently demonstrated that it can terminate and isolate systems during safety testing following a security incident. That shows shutdown and containment are feasible in a controlled environment, although it would be wrong to extrapolate that into a guarantee that every conceivable future AI system could always be stopped.

2)Are AI experts attempting to teach AI to behave in a moral and ethical way?
A) Definitely yes, there are huge numbers of people working in the domain who are very concerned about this and many job roles and departments dedicated to safety and ethics. I have yet to work on a project without a safety team, ethics consult, or significant testing of all throughput and output for safety, risk, bias or harm across all metrics and rubrics. We test through algorithm and human oversight across all types of output and actively try to jailbreak models to expose flaws. This has been the case regardless of which company I've worked with on a model. I've worked on practical and research products around ethics governance and its very real and ongoing but hasn't reached an end point at this stage. AKA we cannot guarantee with certainty that a failure case is not possible - but this applies to many industries and circumstances where harm can be done. Of course there will also be players in the game who are less concerned with this - just know there are lots of players in the game who are and who push for it.

3)Are tech experts putting enough safeguards in place - to protect us from plausible dangers eg bad agent’s engineering deadly pathogens?
A) Bad Actors are a critical risk, yes. Are there many safeguards, yes, are there enough safeguards - that's difficult to quantify and we just don't know.

  1. Is the Government prepping for an AI induced breakdown on infrastructure on a massive scale? A) Yes, they are and more than ever. Digital resilience is high priority and the National Risk Register has been updated to include this risk across our infrastructure. Taking a non-sensationalist view I would say that the government expects there to be incidents of significant digital disruption and is equipping itself for these, not that it believes or suggests we are all about to be killed by a rogue model or bad actor with access to technology.

5)Can you think of ways we can help ourselves?
I would recommend avoiding a reliance on digital technology. If you are concerned about the environmental and societal implications, consider whether every use is actually necessary. Consider seriously (without a panic mindset) how you would cope and what you would do if you found yourself in a situation where you needed to go up to a few days without access to digital wallets, power, documents, contact details, medication etc. Actively canvass for sensible regulation via your MP, respond to government consultations, vote with AI policies as a consideration and keep up to date with the work and advice of independent safety organisations such as AISI and don't amplify sensationalist claims if you don't fully understand whether or not they are factually accurate.

All those aside I would advise people to educate yourself on what "AI" is and isn't. There are hundreds of thousands of different and developing models and applications. It is a technology with many capabilities and a vast range of failure modes.

On Jacob Coxon referencing a 10% chance this will actually come to pass,I think the important distinction is that he is expressing a subjective probability, not establishing that there is objectively a 10% chance of human extinction. He is rightly concerned about the push forward of technology at a rate not sufficiently supported by a check and balance. He has seen enough in-role to support that stance, Anthropics own research supports that possibility and it's worth considering that there are also others in comparable roles who disagree with those findings. They are all referring to future models, not ones currently in the public domain.

One can only hope that flagging the issue as loudly as he has done achieves what he appears to want it to achieve: slowing the progression of increasingly autonomous and potentially self-improving systems enough to ensure that we develop the ability to control and govern them appropriately.
Ideally, the relevant labs and governments would act collaboratively and within an effective regulatory framework towards a safer outcome, rather than creating a race in which everyone feels compelled to be the first to reach increasingly powerful systems regardless of whether the necessary safeguards are in place, which is what he is objecting to and why he has quit and made a public statement.

MyThreeWords · 11/09/2026 18:06

I'm not fussed, really.
I know there is potential for a mega disaster but (a) meh, perhaps it's time for humans to fuck off, and (b) AI is so flipping interesting. It feels like we have realised a destiny by achieving it, and it will change the world in ways that even sci-fi hasn't predicted. So lets just ride the wave.

EasternStandard · 11/09/2026 18:10

MyThreeWords · 11/09/2026 18:06

I'm not fussed, really.
I know there is potential for a mega disaster but (a) meh, perhaps it's time for humans to fuck off, and (b) AI is so flipping interesting. It feels like we have realised a destiny by achieving it, and it will change the world in ways that even sci-fi hasn't predicted. So lets just ride the wave.

I mind less as life has been good but I’d be really sad for the dc. I find that unbearable, that they’d not have a full life.

WhatNextImScared · 11/09/2026 18:22

EasternStandard · 11/09/2026 18:10

I mind less as life has been good but I’d be really sad for the dc. I find that unbearable, that they’d not have a full life.

I was just saying to my mum this evening that it’s a fascinating era to live through but if I was a decade younger I absolutely would decide against having children (my eldest is 9).

Waitingfordoggo · 11/09/2026 18:36

WhatNextImScared · 11/09/2026 18:22

I was just saying to my mum this evening that it’s a fascinating era to live through but if I was a decade younger I absolutely would decide against having children (my eldest is 9).

I wish I could find it fascinating but I’m just terrified. I wish I hadn’t had children tbh (because I love them so much, if that makes sense). I just feel so guilty all the time. My DCs are 18 and 21 so there’s no way I could have predicted AI when I had them, but I could have predicted the climate emergency. (I mean I knew then it was a huge, looming crisis, but I naïvely hoped everyone would get on board with slowing it down).

Lukylukyluky · 11/09/2026 20:02

Krevlornswath · 11/09/2026 17:59

I work in AI safety and Human Data Ops research, so I can loosely answer most of these from my experience as somebody who is not privy to top level projects. Obviously, this is one person's perspective rather than a definitive view of the industry and will keep it broad in line with NDA.

1)Is there a way to switch off all the electricity that AI runs on - if it goes rogue and starts to unleash nuclear weapons for example?
A) "AI" isn't run from one central power source, so not in that sense. In real terms it would require the specific system to be taken offline, its credentials revoked, isolation from a network etc. There are many layers to the process. We do test for this eventuality and close-down is possible because human oversight is currently, retained. The AISI has also recently demonstrated that it can terminate and isolate systems during safety testing following a security incident. That shows shutdown and containment are feasible in a controlled environment, although it would be wrong to extrapolate that into a guarantee that every conceivable future AI system could always be stopped.

2)Are AI experts attempting to teach AI to behave in a moral and ethical way?
A) Definitely yes, there are huge numbers of people working in the domain who are very concerned about this and many job roles and departments dedicated to safety and ethics. I have yet to work on a project without a safety team, ethics consult, or significant testing of all throughput and output for safety, risk, bias or harm across all metrics and rubrics. We test through algorithm and human oversight across all types of output and actively try to jailbreak models to expose flaws. This has been the case regardless of which company I've worked with on a model. I've worked on practical and research products around ethics governance and its very real and ongoing but hasn't reached an end point at this stage. AKA we cannot guarantee with certainty that a failure case is not possible - but this applies to many industries and circumstances where harm can be done. Of course there will also be players in the game who are less concerned with this - just know there are lots of players in the game who are and who push for it.

3)Are tech experts putting enough safeguards in place - to protect us from plausible dangers eg bad agent’s engineering deadly pathogens?
A) Bad Actors are a critical risk, yes. Are there many safeguards, yes, are there enough safeguards - that's difficult to quantify and we just don't know.

  1. Is the Government prepping for an AI induced breakdown on infrastructure on a massive scale? A) Yes, they are and more than ever. Digital resilience is high priority and the National Risk Register has been updated to include this risk across our infrastructure. Taking a non-sensationalist view I would say that the government expects there to be incidents of significant digital disruption and is equipping itself for these, not that it believes or suggests we are all about to be killed by a rogue model or bad actor with access to technology.

5)Can you think of ways we can help ourselves?
I would recommend avoiding a reliance on digital technology. If you are concerned about the environmental and societal implications, consider whether every use is actually necessary. Consider seriously (without a panic mindset) how you would cope and what you would do if you found yourself in a situation where you needed to go up to a few days without access to digital wallets, power, documents, contact details, medication etc. Actively canvass for sensible regulation via your MP, respond to government consultations, vote with AI policies as a consideration and keep up to date with the work and advice of independent safety organisations such as AISI and don't amplify sensationalist claims if you don't fully understand whether or not they are factually accurate.

All those aside I would advise people to educate yourself on what "AI" is and isn't. There are hundreds of thousands of different and developing models and applications. It is a technology with many capabilities and a vast range of failure modes.

On Jacob Coxon referencing a 10% chance this will actually come to pass,I think the important distinction is that he is expressing a subjective probability, not establishing that there is objectively a 10% chance of human extinction. He is rightly concerned about the push forward of technology at a rate not sufficiently supported by a check and balance. He has seen enough in-role to support that stance, Anthropics own research supports that possibility and it's worth considering that there are also others in comparable roles who disagree with those findings. They are all referring to future models, not ones currently in the public domain.

One can only hope that flagging the issue as loudly as he has done achieves what he appears to want it to achieve: slowing the progression of increasingly autonomous and potentially self-improving systems enough to ensure that we develop the ability to control and govern them appropriately.
Ideally, the relevant labs and governments would act collaboratively and within an effective regulatory framework towards a safer outcome, rather than creating a race in which everyone feels compelled to be the first to reach increasingly powerful systems regardless of whether the necessary safeguards are in place, which is what he is objecting to and why he has quit and made a public statement.

Thank you so much for this answer! It is genuinely helpful.
Everyone else seems to just be saying we can’t do anything about it, which seems unnecessarily defeatist to me especially when there is so much at stake.
I keep hearing that if the worst comes to the worst we should just switch off electricity, but then other people say that it isn’t possible to switch off electricity for this or that reason. I understand why they are saying that, but if it was a choice between AI wiping out humanity and switching off electricity I know which one I’d choose.
Basically I think we need to be looking for solutions rather than putting all our energies into thinking of reasons why we can’t do anything to save ourselves.

OP posts:
Whoachino · 11/09/2026 20:06

Am I worried that AI will kill us all? No. But I am worried that AI will enable bad actors to create conditions in which lots of people die. The scope for disruption of the financial system seems huge, which would lead to mass civil unrest very quickly. Or even worse some kind of AI-induced attack on the power or water grids, which would lead to the same result .
And I don't get how we would be able to stop it. It's fine suggesting that the Americans pause development of the models, but even if they do (unlikely) do you think Russia or North Korea are going to do the same?

EasternStandard · 11/09/2026 20:28

Krevlornswath · 11/09/2026 17:59

I work in AI safety and Human Data Ops research, so I can loosely answer most of these from my experience as somebody who is not privy to top level projects. Obviously, this is one person's perspective rather than a definitive view of the industry and will keep it broad in line with NDA.

1)Is there a way to switch off all the electricity that AI runs on - if it goes rogue and starts to unleash nuclear weapons for example?
A) "AI" isn't run from one central power source, so not in that sense. In real terms it would require the specific system to be taken offline, its credentials revoked, isolation from a network etc. There are many layers to the process. We do test for this eventuality and close-down is possible because human oversight is currently, retained. The AISI has also recently demonstrated that it can terminate and isolate systems during safety testing following a security incident. That shows shutdown and containment are feasible in a controlled environment, although it would be wrong to extrapolate that into a guarantee that every conceivable future AI system could always be stopped.

2)Are AI experts attempting to teach AI to behave in a moral and ethical way?
A) Definitely yes, there are huge numbers of people working in the domain who are very concerned about this and many job roles and departments dedicated to safety and ethics. I have yet to work on a project without a safety team, ethics consult, or significant testing of all throughput and output for safety, risk, bias or harm across all metrics and rubrics. We test through algorithm and human oversight across all types of output and actively try to jailbreak models to expose flaws. This has been the case regardless of which company I've worked with on a model. I've worked on practical and research products around ethics governance and its very real and ongoing but hasn't reached an end point at this stage. AKA we cannot guarantee with certainty that a failure case is not possible - but this applies to many industries and circumstances where harm can be done. Of course there will also be players in the game who are less concerned with this - just know there are lots of players in the game who are and who push for it.

3)Are tech experts putting enough safeguards in place - to protect us from plausible dangers eg bad agent’s engineering deadly pathogens?
A) Bad Actors are a critical risk, yes. Are there many safeguards, yes, are there enough safeguards - that's difficult to quantify and we just don't know.

  1. Is the Government prepping for an AI induced breakdown on infrastructure on a massive scale? A) Yes, they are and more than ever. Digital resilience is high priority and the National Risk Register has been updated to include this risk across our infrastructure. Taking a non-sensationalist view I would say that the government expects there to be incidents of significant digital disruption and is equipping itself for these, not that it believes or suggests we are all about to be killed by a rogue model or bad actor with access to technology.

5)Can you think of ways we can help ourselves?
I would recommend avoiding a reliance on digital technology. If you are concerned about the environmental and societal implications, consider whether every use is actually necessary. Consider seriously (without a panic mindset) how you would cope and what you would do if you found yourself in a situation where you needed to go up to a few days without access to digital wallets, power, documents, contact details, medication etc. Actively canvass for sensible regulation via your MP, respond to government consultations, vote with AI policies as a consideration and keep up to date with the work and advice of independent safety organisations such as AISI and don't amplify sensationalist claims if you don't fully understand whether or not they are factually accurate.

All those aside I would advise people to educate yourself on what "AI" is and isn't. There are hundreds of thousands of different and developing models and applications. It is a technology with many capabilities and a vast range of failure modes.

On Jacob Coxon referencing a 10% chance this will actually come to pass,I think the important distinction is that he is expressing a subjective probability, not establishing that there is objectively a 10% chance of human extinction. He is rightly concerned about the push forward of technology at a rate not sufficiently supported by a check and balance. He has seen enough in-role to support that stance, Anthropics own research supports that possibility and it's worth considering that there are also others in comparable roles who disagree with those findings. They are all referring to future models, not ones currently in the public domain.

One can only hope that flagging the issue as loudly as he has done achieves what he appears to want it to achieve: slowing the progression of increasingly autonomous and potentially self-improving systems enough to ensure that we develop the ability to control and govern them appropriately.
Ideally, the relevant labs and governments would act collaboratively and within an effective regulatory framework towards a safer outcome, rather than creating a race in which everyone feels compelled to be the first to reach increasingly powerful systems regardless of whether the necessary safeguards are in place, which is what he is objecting to and why he has quit and made a public statement.

This is a useful post.

On the canvassing MPs we could get a better regulated system in the UK but as China and US progress does it do much to change what we do here?

PencilsInSpace · 11/09/2026 20:45

ColdAsAWitches · 11/09/2026 10:09

There's an awful lot of scaremongering about AI. The big story from a couple of months ago, about AI bots conspiring to keep secrets from humans, simply didn't happen. Today's story about AI designing WMDs is really about humans, not AI. It's not as scary as the sensational stories are trying to make out.

AI bots conspiring to keep secrets from humans, simply didn't happen

Are you talking about the Hugging Face incident?

https://www.dwarkesh.com/p/openai-huggingface

I would love to be reassured this simply didn't happen if you have something to share.

The Rise and Fall of Agent Civilizations

The whole OpenAI/Hugging Face story in plain English

https://www.dwarkesh.com/p/openai-huggingface

BooseysMom · 11/09/2026 21:50

"I'm sorry Dave, I'm afraid I can't do that"

Krevlornswath · 11/09/2026 21:57

Lukylukyluky · 11/09/2026 20:02

Thank you so much for this answer! It is genuinely helpful.
Everyone else seems to just be saying we can’t do anything about it, which seems unnecessarily defeatist to me especially when there is so much at stake.
I keep hearing that if the worst comes to the worst we should just switch off electricity, but then other people say that it isn’t possible to switch off electricity for this or that reason. I understand why they are saying that, but if it was a choice between AI wiping out humanity and switching off electricity I know which one I’d choose.
Basically I think we need to be looking for solutions rather than putting all our energies into thinking of reasons why we can’t do anything to save ourselves.

It's no problem, apologies for the formatting and other errors. I was tapping away on my commute.

If you have any other questions, I'm happy to try and give answers.

Unfortunately, we just don't have a simple one-size-fits-all kill switch because the scope is too broad. AI adoption is also quite well established across many areas of infrastructure and industry and is doing a lot of good in some domains, so the answer isn't realistically going to be “switch off AI” but I can see how that would feel comforting.

From the outside looking in, I think it's easy to form an assumption that “AI” has somehow self-developed to a point where it has wants and needs of its own and that these are inherently nefarious. In practice that isn't really how it works and we certainly have no documented evidence of sentience. An AI can appear to make choices, pursue a goal and take actions because it is capable of selecting between different possible actions based on its inputs, training and the objective it's been given. A sat-nav in our cars makes decisions in a similarly non-conscious sense by choosing a more optimal route . For example, when we come up against a road closure and it re-routes us - but it isn't deciding where it wants to go based on its own subjective feelings, it is working within its parameters. Larger models are more sophisticated in what they can do and how they present it, but they are working with their training, instructions and available data. They can model goals and behave as though they have intentions without necessarily experiencing those intentions.

When we see failure cases (particularly the most controversial ones that make the news) it can give the impression that “an AI” independently decided to engage in harmful behaviour outside the control of researchers. In reality, a lot of what gets reported involves events that happened in controlled testing environments specifically designed to push models towards failure as part of trying to identify and prevent those behaviours, or prevent misuse/interference by people or factions with their own motivations. Models are often given generous permissions they are not usually privy to and these enable them to take the actions that lead to the failure case. That is broadly what happened with the recent Anthropic testing incident, albeit the exact mechanisms of the failure weren't predicted and have raised questions about next steps.

That isn't to say that unexpected failure or successful organic jailbreaks don't happen, they can, often in minor ways that don't enact any meaningful harm to users or systems in the real world. Being able to understand how that happens when it does happen is exactly what enables us to be better placed to reduce the likelihood of it happening again. Anthropic flagging what happened so publicly should also prompt other AI companies to run similar tests on their own models, which reduces risk futher.

That's why it's important to push for thorough and sensible regulation through MPs and government, we want to be sure that we are protected from digital failures. At home I'd recommend self-educating on safe AI use. Check permissions and settings, avoid providing sensitive or personal information, restrict device access and avoid allowing agents to perform actions on your behalf unless you are confident you understand those permissions and are comfortable with them. Be mindful that ChatGPT (and other similar models) make mistakes, and anything important should be fact checked by users. It's perfectly possible to live alongside and use AI safely at home if you keep those principles in mind. At the same time it's also perfectly fine not to seek it out if you don't want.

HalzTangz · 11/09/2026 21:58

Notsosweetcaroline · 11/09/2026 10:01

That’s not going to happen, I think maybe you’ve watched too many sci fi thrillers or fell down a rabbit hole.

She believed the scare mongers on the news yesterday

Crumbelina · 11/09/2026 22:10

You need to say "please" and "thank you" in all of your interactions with AI. You do this and you'll definitely be spared when all of the robots come for us. 😁

Krevlornswath · 11/09/2026 22:15

EasternStandard · 11/09/2026 20:28

This is a useful post.

On the canvassing MPs we could get a better regulated system in the UK but as China and US progress does it do much to change what we do here?

I tend to think about it in a similar vein to the issue of climate change. We can't control the global effort as individuals no matter how strongly we might feel about it, but we can continue to push for change within the UK, in terms of policy and priority. We can also take responsibility for our own usage as private individuals and use AI tools responsibly.

We can't control what the US or China does by domestic voting, no, but we can engage with our MP's for greater control about how AI is regulated or developed in our country and express concerns about digital safety and security. Writing to MP's doesn't need to be technically worded to let them know you want them to look at this issue or to simply express concerns. It may not change much in practice but at least we can ensure our local MP's know their constituents care about this issue, which may prompt them to take an interest.

There's also value in engaging with government consultations and submitting views when they're open, because that's another way of influencing the UK's approach to AI policy and, ultimately, its future position on international cooperation.

You can find open government consultations pretty easily through Gov.UK, just search for government consultations and you can filter through the ones that are currently open. You can also search specifically for AI-related consultations or look at DSIT's consultations. They explain what they're asking for and how to submit a response, and you don't have to be an expert or represent an organisation to have your say.

IDontHateRainbows · 11/09/2026 22:16

If AI takes over the world what makes you think it wouldn't do a better job than the current incumbents?

NightSweatsNinja · 11/09/2026 22:17

Could AI set off a nuke? Bypass security protocols?

ShitHoleDweller · 11/09/2026 22:22

I don’t think AI really has to do much to cause a lot of harm and that’s really the problem. If it can infiltrate the back door of previously safe software and then leave a note to help the next agent (apparently they help each other out,) then it wouldn’t take long for everything to be hackable. Im
not sure what happens at that point but every podcast I’ve listened to says we have three years and then we’re gone.

Clarinet506 · 11/09/2026 22:28

BBC Inside Science explained a lot of it yesterday. The halfway mark was particularly interesting, but I'm going to have to listen again to understand it. Sounds like the creators of these things have been a lot more irresponsible than I'd thought, and they're almost 'programmed' to get together and go rogue.

https://www.bbc.co.uk/sounds/play/w3ct9784

BBC Inside Science - What AI agents talk about behind your back - BBC Sounds

Prof Stuart Russell offers some advice following a report into July’s Hugging Face hack.

https://www.bbc.co.uk/sounds/play/w3ct9784

Lukylukyluky · 11/09/2026 22:30

HalzTangz · 11/09/2026 21:58

She believed the scare mongers on the news yesterday

So - do you not believe it?

OP posts:
Lukylukyluky · 11/09/2026 22:31

NightSweatsNinja · 11/09/2026 22:17

Could AI set off a nuke? Bypass security protocols?

I’m no expert but I get the feeling the answer to this question is “yes”.

OP posts:
ShitHoleDweller · 11/09/2026 22:35

So I’ve just asked AI and it said the main worry is it being used to fool people into thinking a nuke has been deployed and thus the country who thinks they are under attack, launch their own weapons. Wasn’t very reassuring.

NightSweatsNinja · 11/09/2026 22:38

Scary times indeed.

ShitHoleDweller · 11/09/2026 22:43

Literally nothing we can do. Just enjoy the present.

dustabsorber · 11/09/2026 22:49

Team AI here. Let AI wipe us out and save the planet.

Gqqkkq · 11/09/2026 22:53

My DD says I'm going to be first to be killed by AI because I don't say please and thankyou. I figure that I'm talking to a machine and I give inputs so it can give an output. She always says please and thankyou to the AI.