Anthropic's proposed AI safety bill only allows government to block models under extreme conditions
Channel: CBS News
Uploaded:
Duration: PT335S
Read a transcript preview of “Anthropic's proposed AI safety bill only allows government to block models under extreme conditions” from CBS News, uploaded on 2026-10-22.
Transcript preview
Language: English (original, YouTube automatic captions)
[0:00] Welcome back to Takeout. We are learning
[0:02] a bit more about the draft of an AI
[0:04] safety bill that Anthropic quietly
[0:06] pitched to lawmakers on Capitol Hill. It
[0:08] would allow the federal government to
[0:10] block artificial intelligence tools and
[0:12] models.
[0:14] Brace yourself. Only if they kill or
[0:17] injure at least 50 people or inflict at
[0:20] least $1 billion in damages. That's what
[0:23] the bill actually says. president
[0:24] closely follows California's legislation
[0:26] which allows the state to regulate
[0:29] developers of so-called frontier AI
[0:31] models. Joining me now Nate Sores is
[0:33] executive director of the machine
[0:35] intelligence research institute and he's
[0:37] the co-author of the book if anyone
[0:39] builds it everyone dies while why
[0:42] superhuman AI would kill us all. And I
[0:44] know that's a scary title.
[0:46] Nate does not mean to do this does not
[0:48] need me to do this. I get no residuals,
[0:50] but if you are curious about this space,
[0:52] that's a great book to read to intr