“I'm not sure that's the best way because … if we go down that path as we are, there will be incidents that happen and hopefully they're not too awful, but you don't know. And then there'll be regulation. … We tend to wait till something happens. And then think of the global financial crisis 2008 and then a whole bunch of new regulations come in. So that's the likely case here, unfortunately.”
Masters of Scale · Startups & Venture · October 2026
Context
Swisher asked about AI companies policing themselves without guardrails or regulation. Hastings, who sits on Anthropic's board, says waiting for a disaster before regulating is the likely path, and predicts a 2008-style rush of rules after an incident.
Transcript
…gotten much better just in the last 18 months with a thing called classifiers, which is like a front end on your AI usage. So if you ask it, you know, how do I rebuild a COVID-19, you know, it doesn't let you do it because that's this separate subsystem, the classifiers. It's not perfect. But right
…for lunch, it seems like, which is every Cenan always calls, he's like, tech people are at the White House. I'm like, call me when they're not, actually. That's what I'd like to comment on. But self-policing, self-policing, no guardrails, self-policing, no regulation. How do you look at that?
I'm not sure that's the best way because I think this, you know, if we go down that path as we are, there will be incidents that happen and hopefully they're not too awful, but you don't know. And then there'll be regulation. You know, so we tend to wait till something happens. And then think of the global financial crisis 2008 and then a whole bunch of new regulations come in. So that's the likely case here, unfortunately. Might
it be a better idea to possibly anticipate those things like you do every 10 minutes?
You should be at the White House. I feel
like I'm not going to get invited to the White House. Crazy.
More from Masters of Scale
