John Kennedy tries to fast-track a bill requiring AI companies to have a kill switch
United States senator from Louisiana, quoted in an ABC News report
“If there is even a 1% chance that one of these AI models can shed its nature as a tool and become an independent species.”
Source and context
Original post
About this source
ABC News segment posted to the network's X account at 11:39 AM ET on September 17, 2026. Besides Hinton it carries several other figures relevant to this case and not yet recorded from it: Sen. John Kennedy attempting to fast-track a bill requiring AI companies to be able to deploy a kill switch, saying there is "even a 1% chance" a model could shed its nature as a tool and become an independent species; Sen. Rand Paul blocking that proposal on the grounds that Congress does not understand the technology well enough to regulate it; and an unnamed senior Trump adviser saying existing law already makes AI companies liable for harmful products. The segment also reports that Anthropic's chief executive wants coordinated industry safety standards and has said this may require an antitrust waiver, which senators of both parties have resisted. One line in the segment — "This is bat crazy insane science fiction stuff" — is spoken over Hinton's probability estimate and is not Hinton's; the speaker is not identified in the clip, so it is not attributed here.
Before the quotation
Geoffrey Hinton had warned Congress it has roughly a year to act, and senators of both parties had resisted Anthropic's suggestion that coordinated industry safety standards might require an antitrust waiver.
After the quotation
Rand Paul blocked the proposal, saying Congress does not understand the technology well enough to regulate it, recorded separately in this case.
How this statement is classified
Case context: Should Washington impose stronger safeguards on frontier AI?
The label describes this statement’s response within the context above.
Why this label?
Kennedy supports it concretely, seeking to pass a federal mandate that AI companies be capable of shutting their systems down, and justifying it on a low-probability, high-consequence basis. That places him with Cantwell (728) and Hawley (798), who press for specific federal requirements, rather than with Huang (727) or Johnson (729). The closest competing label is “Mixed or conditional”; it loses because he attaches no condition to the requirement itself. Note that Kennedy questioned Kash Patel about FBI hiring standards in a separate case; this record concerns only his AI legislation.
- Recorded on
- Published here
Should Washington impose stronger safeguards on frontier AI?
Explore the case context, sources and public responses.
More from this case
Read the full caseFuture of Life Institute
“If they want to protect their citizens, every country must immediately implement and enforce: Licensing regimes that force AI companies to guarantee that systems are controllable and safe, both pre-development and pre-release. Liability for AI companies in general and criminal liability for executives if their systems cause significant harm. A halt to the development of the most dangerous capabilities, in particular recursive self-improvement. Exclusion of companies from their markets if those companies cannot comply with basic safety requirements.”Read statement
“Put more simply, the models are increasingly smart enough. They know when we’re watching them, and they change their behavior accordingly. So what they do when we are testing them, when we audit them, may not tell us what they’ll do in the wild. It surprises me that this counts as a radical proposal, but here it is: If you are losing your ability to evaluate the models you have now, maybe don’t let them build models you’ll be even less capable of controlling in the future. Important from @ezraklein”Read statement
OpenAI
“That is why we believe the United States should lead an effort to work together with countries around the world to develop global technical standards for frontier AI, including for RSI.”Read statement