Distinguished Teaching Professor, School of Law; formerly Assistant Attorney General for Antitrust, U.S. Department of Justice, Washington University in St. Louis
Jonathan Kanter, former assistant attorney general for the Antitrust division, joins 'Squawk on the Street' to discuss Trump's newly formed 'Super Intelligence' force, the future regulation of AI, and more.
What Jonathan Kanter said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Jonathan Kanter criticized the new 'super intelligence force' as 'vaporware,' arguing that non-binding principles are ineffective without consequences. He stressed the need for basic rules and duties of care, distinguishing them from heavy regulation, and rejected Jensen Huang's claim that existing liability laws suffice, citing that it took over a decade to hold social media accountable. Kanter denied that AI labs need blanket antitrust immunity for safety cooperation, warning against immunity based on the Section 230 precedent. He advocated for a 'dog bites neighbor' rule holding companies responsible for AI agents' actions, and acknowledged that current product liability law is uncertain and slow, urging Congress to act, despite its paralysis. He also predicted accountability would become a central issue in upcoming elections.
Key takeaways
Kanter dismissed the super intelligence force as 'vaporware,' saying non-binding principles are ineffective.
Kanter called for a 'dog bites neighbor' rule, holding companies liable for AI agents' harmful actions.
Kanter said Congress is 'at a point of paralysis' and must create basic duties of care.
0:10What does it mean for the regulation of AI companies as they clamor for antitrust exemptions?
1:19What's your response to those who would say, you know, you're standing in the way of progress?
2:02Now, leaders in AI such as Jensen Huang say the liability laws already on the books in our country in various states will be enough that they will prevent companies from doing things that are unsafe. What do you say?
3:05government as part of the DOJ, the labs say that, or at least we've been hearing this idea that they need so-called blanket antitrust immunity to be able to fully cooperate on safety. Is that the case?
4:11For example, what are the liabilities to a company such as OpenAI for the damage that might be caused?
6:20Can you not make that case that because they developed this model and they cause whatever damage it may be, could be physical damage and maybe otherwise, that they could be held liable? Are laws insufficient?
7:09I mean, even if the Democrats were to win the Senate and the House, uncertain obviously, a month from now in the midterms, do you expect that that would really change what you just described?
Welcome back. President Trump announcing members of a new super intelligence force, which he says will coordinate the government's engagement with AI companies, consumers and infrastructure providers. It will be led by Director of National Intelligence Jay Clayton. It also includes FTC Chair Andrew Ferguson, Undersecretary of Defense Emil Michael and Director of the Office of Personnel Management Scott Kapor. What does it mean? What does it mean for the regulation of AI companies as they clamor for antitrust exemptions? Well, let's bring in somebody who knows about that. Former Assistant Attorney General for Antitrust, Jonathan Kanter. Jonathan, always good to have you. I guess first, before we get to antitrust, just give me your overall take on this new body, so to speak, that we just mentioned. The members of that are going to be overseeing with what we expect will be a very light regulatory touch. AI regulation.
It's basically vaporware. It's nice to have principles, but non-binding principles are like traffic lights being a suggestion rather than a rule. And so if you only have to stop at the red light when you want to, and there are no consequences if you blow through it, it really doesn't do anything.
All right. I can hear some of you saying, yeah, just another Democratic obstructionist. There he is. All he wants is regulation. What's your response to those who would say, you know, you're standing in the way of progress?
No, I think we want lines on the road and stop signs and traffic lights. We just want some basic rules. There's a difference between rules and regulations. We want to make sure that there are standard, basic duties of care so that if companies violate those duties of care they're held responsible and that shareholders are going to hold the companies' feet to the fire to make sure they comply. We're inventing technology that is transformative. It could have major impact on how they do major good, but it could also do major harm and the public is saying we want some basic safety and security.
Yeah. Now, leaders in AI such as Jensen Huang say the liability laws already on the books in our country in various states will be enough that they will prevent companies from doing things that are unsafe. What do you say?
I think they're helpful but not sufficient. So products liability has ultimately been a useful framework or at least an effective, recently effective framework against social media. But it took us over a decade to get there. I think we need to establish basic principles and duty of care that essentially say, if you build harmful products and they do bad things, you're held responsible. And the consequences of violating the duty of care have to be significant enough so that the companies have an incentive to comply. What I say is not that this is over-regulatory, it's that it's not fair to put this on the shoulders of the companies. They have fiduciary obligations. They need to innovate. We need to give incentives to make sure that they innovate in the right way, to make sure that the products are both useful and safe, and that will result in us having better AI, better products, and better productivity.
Let me come to you now as an expert on antitrust. Having once run that for the U.S. government as part of the DOJ, the labs say that, or at least we've been hearing this idea that they need so-called blanket antitrust immunity to be able to fully cooperate on safety. Is that the case?
No, they don't. So the labs do not need blanket antitrust immunity to cooperate in safety. Existing laws, properly structured collaborations on standards, sharing information about malicious actors and threats are all perfectly permissible under the current laws. So no immunity is necessary. And, you know, I, for one, am a little bit concerned or suspicious of any immunity in tech. I think we saw the results of Section 230, which conferred immunity early on for a very narrow purpose that ended up becoming a liability shield, that allowed social media companies to essentially become invulnerable to the consequences of building bad products. And so I think we should be very careful about imposing antitrust immunity, especially when they can collaborate in safety and security without one.
8 more exchanges in this transcript
Sign in free to read the rest of this interview. No card required.
Kanter, J. (2026, October 5). Watch CNBC's full interview with former assistant attorney general, Jonathan Kanter [Interview transcript]. CNBC Television. CEOInterviews.AI. https://ceointerviews.ai/interview/3094819/
MLA
Jonathan Kanter. "Watch CNBC's full interview with former assistant attorney general, Jonathan Kanter." CNBC Television, 5 Oct. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3094819/.
BibTeX
@misc{kanter2026_3094819,
author = {Jonathan Kanter},
title = {Watch CNBC's full interview with former assistant attorney general, Jonathan Kanter},
howpublished = {Interview transcript, CNBC Television. CEOInterviews.AI},
year = {2026},
month = {oct},
url = {https://ceointerviews.ai/interview/3094819/},
note = {Speaker-attributed transcript with timestamps}
}
