Organizing an AI Safety Summit

Maybe this is too ambitious / crazy, but it also seems like we may actually be able to pull it off: organizing an AI Safety Summit. Because if no-one is being the adult in the room, perhaps we need to be the adults.

I've been focused on a summit since day one, but the focus always was on convincing someone to organize a summit. The UK did, but they fucked up in a couple of important ways: no treaty, no international org, and worst of all: they passed on the stick to France. And as you probably know by now, they gutted the november AI safety summit (postponed it, removed safety focus, led by AI-skeptic).

So maybe we need to play a more active role in all of this. Shape the agenda in a meaningful way. Get the right people to visit. Make the right preparations (e.g. treaty draft).

What would it take


  • A suitable location
  • A meaningful programme (with big name speakers)
  • Buy-in from important attendees
  • Money (at least... 100k?)
  • Website
  • Press releases
## Questions

  • How do we make this legitimate? How do we get the big names to join? I think getting the big AI safety names is not a big challenge. The difficulty is getting the countries to join. Branding it as "pauseAI" is probably not the way to go.
  • What is our invite strategy? Do we send out invites to a couple of big names first and use these for bragging rights?
  • Who can fund this? How much will it even cost?
  • Who do we really need to talk to about this?
## How you can help

  • (easy) Do research in how earlier summits were organized. Share learnings below!
  • (medium) Get us in touch with the right people
  • (hard) Help us fund this thing
  • (nightmare mode) Make it happen. Dedicate hours. Commit to realising this.
Was this page helpful?