Possible mistake EAs are making and shout out to Pause AI UK

By Michelle_Hutchinson @ 2026-06-24T20:22 (+146)

I think right now EAs might be making a significant mistake by paying insufficient attention to the political realm. As EAs we tend to figure out what’s most impactful for us to work on and focus hard. That’s great! But there are various actions that are ‘non-delegatable’ - the extent to which an individual can do the action is limited (like voting, going to a protest, making hard money contributions to particular campaigns). It might be useful if we were all more in the habit of doing various of these alongside what we’re most focused on. I think more attention is starting to go to this - as evidenced by Jeff’s blog post about how he thinks political donations should be his primary method of giving going forward. But I think we probably have a ways to go.  

Political actions look better than they used to

I think it’s pretty unsurprising that EAs are sceptical of lobbying policy makers as a theory of change. By my lights, it looked decidedly less effective a decade ago. At that point, work on existential risk was very speculative, and mostly needed additional research. It was totally unclear what better legislation would look like. Meanwhile, global poverty and animal welfare were (and are) severely bottlenecked on money. As an individual in one of the richest countries in the world, if you’re willing to give 10% of your income, that’s a lot of money. It seems hard to get to similar expected value by political actions aimed at increasing government development spending. 

From the point of view of existential risks, we’re now in a very different world. We have a better sense of what risks we might face, and of what types of legislation will in expectation improve the safety of the world. Yet it’s far from clear that that legislation will be implemented. 

Knowing it’s important to have the right legislation is a far cry from knowing what that legislation is or how to engage with the democratic process in a way that will be helpful, mind. Personally, I’m particularly wary of people-engaging things, because I’m more at home in googledocs, and rather worry I’ll misstep and accidentally make things worse. 

An example of making it tractable: PauseAI UK’s campaign this week

I’ve been grateful recently that PauseAI UK has been doing the work to make this much more tractable for individuals. For example at the moment they’re organising a campaign for the UK to introduce legislation that holds AI companies liable for damages their models cause. The campaign features an open letter calling on the Prime Minister to support such legislation, and a policy briefing as background. They asked supporters of PauseAI to write to their MP, and ask to meet with them earlier this week to discuss signing the letter. 

As someone who’s never written to or met with their MP before, I wouldn’t have known where to start doing so, and it sounded daunting. PauseAI UK actually made the whole process shockingly smooth. The policy briefing and open letter both struck me as well written, reasonable and the right mix of meeting an MP where they’re at and highlighting what’s important. They provided an ‘email builder’ which helps you find out who your MP is, suggests what a letter to them might look like (I liked their draft), and allows you to actually send it to them. I had various questions which Joseph was helpful in answering.

My MP was happy to set a meeting on the day we requested. They met with the 3 of us coming from my constituency in Oxford at the same time. I was expecting just 10 minutes with them, but we actually talked for 45min. PauseAI UK did the leg work of figuring out precisely where we should go, how to get to that entrance, and giving a sense of how these conversations typically go, which made it feel easy to prepare for. 

How valuable and achievable is it for us to do more of this?

It feels difficult to me to know how valuable things like this meeting/campaign are. The MP didn’t commit in the meeting to sign the letter (which makes sense since they hadn’t read it yet). They seemed sympathetic to the idea that guard rails on AI development are important and not yet covered, but often had different particular concerns than me. They said they wanted to follow up afterwards. [Edit: They've now signed the letter!]

Working on things aimed at improving and safeguarding our (crazy) future always seems speculative, making it hard to compare things like talking to my MP with the marginal hours at my day job. To some degree, I try to avoid them trading off against each other. But I am tempted to think that even without fully understanding the value, I ought to spend some time talking to my MP about a what I think the government is likely to get disastrously wrong over the coming few years. 

I’m cognisant that it’s easy for there to be many actions that ‘everyone ought to do to make a democracy function well’ and those end up adding up to more time than anyone has. But I also don’t want to let that reductio prevent me from taking any political actions. After all - improving the governance landscape seems absolutely crucial. I’m keen to hear how others think about the value of these actions and how to prioritise them!

Another thing that makes things tricky in this area is that there are limits to what charities are allowed to say about government. Moreover the rules often aren’t very clear, and talking to lawyers is often expensive and time-consuming. That makes it particularly hard to figure out the effectiveness of different political actions and help each other take the effective ones.  

Thankfully we’re now a large community, brought together by willingness to put quite a bit on the line to help others. If we’re thoughtful and collaborative in how we do that, despite the trickiness I think we can really move the needle in how seriously policy makers take AI progress and its risks.