Hanif questions Anthropic on enforcement in classified military deployments

Councilmember Hanif, followed by Councilmember Aviles, followed by Councilmember Gutierrez.

Thank you, Chair.

On February 28th of this year, 120 schoolchildren and 35 teachers were killed in an airstrike reported as the deadliest U.S.

Attack on civilians in 35 years.

The military has stated the target was misidentified by Project Maven, which originated with Google and Amazon Web Services and later incorporated Palantir software and Anthropix Cloud.

To Mr.

Graham from Anthropic, your usage policy prohibits facilitating any act of violence.

What mechanism enforces that prohibition in a classified deployment, and how would your company learn of a violation?

And has your company ever terminated or suspended a government customer for violating its usage policy?

If so, how many times?

First, I really appreciate this question.

I don't have the exact answers to that.

I can point to our very public positions on...

Could you just share them out loud here?

In government, we've taken very public positions against the use of, for example, domestic mass surveillance, the use of autonomous weapons or development of autonomous weapons.

How does a company learn about that violation?

And then could you also just respond to the piece about the termination of a government customer?

I'm not entirely sure of our process on sort of government sort of monitoring.

I'd refer to my colleagues there.

Thanks.

Thank you so much.