NEW YORK—U.S. Senator Chris Murphy (D-Conn.) on Tuesday joined Foreign Policy magazine’s tech and AI forum at the United Nations General Assembly (UNGA), titled Securing the Stack: A Conversation on Trust, Trade, and Technology. In the panel discussion, Murphy outlined the path ahead for Congress to respond effectively to Americans’ growing concerns around AI, including job loss, competition with China, and humanity losing control over the technology.

Murphy argued that the US has a long history of regulating dangerous new technologies and the American people are demanding action: “It might be the end of our democracy if the people beg on both sides of the aisle for us to do what we've done with every complicated new technology in the history of this country since the industrial revolution – regulate and control it to make sure we get the good but not the bad – if we don't step up and do the work.”

Murphy expressed optimism that lawmakers could find some consensus on AI: “I think there is some low-hanging fruit that is not particularly complicated. The deepfake threat is already real… that's a pretty easy one to deal with, right? You could just outlaw them in politics, or require clear identification or watermarks in other contexts. Friendship bots, as I mentioned before, are poison for our kids. We can outlaw them writ large, or we can outlaw them for kids. And then you know what we're talking about right now, which is some approval process for models, so as to guard against what we are seeing, and potentially [what] we'll see at scale soon as they sort of jump out of containment…”

“Those three things are achievable, not terribly political, and could foreseeably get done, and give the public then some faith that we maybe have the political capacity to handle some of the more nuanced threats down on,” Murphy emphasized.

Murphy stressed that humanity cannot retain its control over AI without cooperation with China: “I don't think there's any path forward without us being in a conversation with the Chinese and the rest of the world on basic AI safety guidelines, the most important of which is maintaining human control over these systems. So it is heartbreaking to me to listen to the president essentially swear off any kind of engagement with the world that would involve any kind of constraint on research in the United States.”

Murphy also called out the Trump Administration’s hypocrisy on AI development: “He’s giving the impression that his priority is to beat the Chinese on the issue of AI. But at the same time he’s signing off on agreements that would have been unthinkable just a year ago to give some of our most sensitive technologies to the Chinese, to basically invite them to lap us in this, in this AI race.”

Murphy warned that failing to regulate AI out of false fear mongering plays right into China’s hands: “I worry about the China red herring blocking us from common sense domestic regulation. China would love nothing more than for the United States to do nothing to protect people spiritually and the economy generally from what AI could do unchecked. They would love for us to be the one that submits our citizens as guinea pigs to this technology, they would love for us to be the ones that have massive job dislocation first. And so, if we want to compete with China, we better get serious about regulating the social and economic dislocation that may happen with AI at speed because they certainly are going to be careful about that.”

Murphy also detailed what role AI companies should play in crafting a regulatory regime: “We shouldn't wait for their permission. We need to do this in the next 12 months. And so, while I would like them at the table, the last thing we should do is give them veto power… I think that would be one of the most dangerous approaches we could take is to only pass regulatory regimes that get the sign-off and approval of these companies. That, by definition, might make it illegitimate in the public's mind, that is already convinced that these companies are not out to protect them, but out to protect their bottom line.”

A full transcript of Murphy’s exchanges is available below. A video of the full discussion is available here.

MODERATOR:

My question to you is, how much do people still trust the United States? And how much do you have to make the case that they do when you have your engagements with the rest of the world?

MURPHY:

Start with an easy one, huh?

Great to be with you all. Thanks for having me. I'm gonna be the representative member of the United States Senate. And I care deeply about this issue, but am far from an expert on it. So I'm gonna leave you the heavy lifting.

I mean, just for a moment, let's just start with the other end of trust: which is the lack of trust that the American people have in their government right now. They've been worrying about AI and its both economic and spiritual impact on the country for years now. Sixty percent of kids in the United States already use chatbots as a sort of regular outlet for friendship and advice, and parents are dealing with the poisonous ramifications of that every single day. And so, the biggest issue of trust is the lack of trust the people have that we're going to actually protect them from the poison, protect them from the potential enormous downside of unregulated, unchecked AI. And it might be the end of our democracy if the people beg on both sides of the aisle for us to do what we've done with every complicated new technology in the history of this country since the industrial revolution–regulate and control it to make sure we get the good but not the bad–if we, if we don't step up and do the work.

No, I mean listen. There's a crisis of trust in America right now. And yes, part of it is at a kind of macro political level in which there's just such whiplash in U.S. policy from administration to administration. But let's just not sugarcoat it. We've never ever had a president like this before: who lies through his teeth, who signs agreements with the world that he has no intention of honoring, that says one thing today and the exact opposite the next day. And so there is a very specific deficit of trust right now.

[At] My meetings [at the UNGA], you're basically with countries that are just trying to survive, that just want to avoid being on the president's radar screen, [that] want to avoid being involved with America. What a disaster for us, that is. But that's the reality of dealing with this very exceptionally dangerous president. So yes, I I worry about the the bigger issue here, which is that you know even if Trump is gone and there's a more kind of mainstream, internationally focused president, there’s just going to be a lot of hesitancy to do a deal with the United States on critical minerals or AI because they worry that will only last for the duration of that particular administration.

MODERATOR:

Just to stay with you for a minute: on the the first part of what you said, on the the fears that the American public right now have around AI, and you said that [they’re] begging government to to do something about it, the conversation around AI has kind of seesawed from the these existential risks that the companies are talking about. But then here at UNGA, there's also a lot of conversation on the more immediate risks, some of which you laid out on impacts on education, job loss. So, on that kind of spectrum of risk, where can Congress most realistically and effectively intervene?

MURPHY:

Well, I think there is some low-hanging fruit that is not particularly complicated.

The deepfake threat is already real. It is not just a political threat that nobody knows whether they're actually hearing true political speech or not. It's hard to describe, but it's kind of a metaphysical threat in that today, when you go online, you have no idea whether what you're watching is real or not. And when you live your entire life not sure of what is truth and what is untruth, it is exhausting. It is spiritually exhausting. And so I just want to speak to the fact that these are not all easy to diagnose threats. They are literally undermining our consciousness on a regular basis. So that's a pretty easy one to deal with, right? You could just outlaw them in politics, or require clear identification or watermarks in other contexts.

Friendship bots, as I mentioned before, are poison for our kids. Poison. We can outlaw them writ large, or we can outlaw them for kids.

And then you know what we're talking about right now, which is some approval process for models, so as to guard against what we are seeing. And potentially, we'll see at scale soon as they sort of jump out of containment. That's more complicated, but not impossible to figure out how to get some third-party eyes and perhaps approvals on new complicated models.

Those three things are achievable, not terribly political, and could foreseeably get done, and give the public then some faith that we maybe have the political capacity to handle some of the more nuanced threats down on.

MODERATOR:

Sticking with the concept of trust, but hopefully linking it a little more to the trade and tech aspects: China. Tomorrow, President Trump and President Xi are meeting in Washington [on] AI technology issues more broadly. Trade issues are expected high on the agenda. You talked about sort of approval process, and then that goes into sort of how much we should pause or slow down development of these frontier models. How much can the U.S. and the rest of the world trust China on that front?

MURPHY:

Well, listen I don't think there's any path forward without us being in a conversation with the Chinese and the rest of the world on basic AI safety guidelines, the most important of which is maintaining human control over these systems. So it is heartbreaking to me to listen to the president essentially swear off any kind of engagement with the world that would involve any kind of constraint on research in the United States.

He’s giving the impression that his priority is to beat the Chinese on the issue of AI. But at the same time he’s signing off on agreements that would have been unthinkable just a year ago to give some of our most sensitive technologies to the Chinese, to basically invite them to lap us in this, in this AI race.

I worry about the China red herring blocking us from common sense domestic regulation. China would love nothing more than for the United States to do nothing to protect people spiritually and the economy generally from what AI could do unchecked. They would love for us to be the one that submits our citizens as guinea pigs to this technology, they would love for us to be the ones that have massive job dislocation first.

And so, if we want to compete with China, we better get serious about regulating the social and economic dislocation that may happen with AI at speed because they certainly are going to be careful about that. So yes, we have to be engaged with them, but we can't wait for that conversation to make some of those basic investments in a regulatory regime that I referenced in answer to your first question.

MODERATOR:

Senator, I want to also talk now about the role of kind of allies and partners - and we are at UNGA - so the role of multilateral institutions and frameworks. On the first one, despite all the kind of upheaval in foreign policy, one of the things that the Trump administration has been very focused on is the Pax-Silica initiative, for example. I'm curious for your thoughts on one multilateral effort that we're seeing come out of the Trump administration. How effective is it? And what more would you like to see in terms of that conversation, of working with partners to secure?

MURPHY:

Yeah, I think it's not going to be terribly effective. Because you know, with one hand he's rubbing the back of our allies, on the other hand, beating them over the head of the club. And so it's really difficult to convince the Europeans, for instance, that they should put money on the table side by side with the Americans to protect global supply chains around critical minerals when we are sanctioning and tariffing them at the same time.

That being said, I actually have been somebody who has argued for our development finance capabilities, which are right now housed at the DFC, the Development Finance Corporation, to be used in a more strategic way. Traditionally, those capabilities were doing more to combat humanitarian problems and attacking poverty and deeply poor nations. Trump has shifted a lot of the DFC's focus into this specific area of critical mineral flows into AI technology, and I don't actually think that there's – he's maybe gone too far – but I think that that's actually a smart move to use our global development finance capabilities to do sort of more geopolitical strategic investment. So I don't think the initiative is going to bear fruit because you know we are just right now defined by the fights we're picking with our allies, and they're just not going to work with us when we're constantly in an antagonistic relationship. They don't know what attack is coming next. But the things that have happened at DFC - there is some silver lining there.

MODERATOR:

And on bodies like the UN, what role do you see for the UN as an institution? Under this president, U.S. engagement with the UN has kind of gone back and forth a little bit, shall we say? But what role do you see for the UN in bringing together and enforcing some of these kinds of governance mechanisms?

MURPHY:

I doubt that the UN is going to be the place where this conversation happens. This is going to be an ad hoc multilateral effort that is going to primarily involve the United States and China and a few other big players in this space.

But just to step back for a minute: what did Trump say this morning? I forget, but he used the word - ‘we're not going to create a global cabal on AI.’ He sort of used that word - global, globalism - and it speaks to sort of the way that Americans feel about our engagement in the world for the past 50 years.

Most Americans believe that our engagement in the world over the last several decades has prioritized the interests of billionaires and corporations, and that our economic integration did not serve them, but served the economic elite. And so, if we are going to pitch any new platform of economic integration to the world, it better lead with a clear explanation about how it's going to help regular people, how it's going to protect the health of local neighborhoods.

And so, I do think people intuitively understand that you probably can't do AI without some integration with the world. But their default belief is that integration and global economic agreement is designed to help the big legacy players and designed to help companies move jobs out of the United States, through labor arbitrage, to wherever place makes them the most money.

So I think it's going to be necessary, but very difficult for any American leader to convince the public that a new global AI framework isn't primarily intended to serve the big companies' interests and their private interests. You're going to have a high burden to explain, as I think we can, about how those agreements internationally are really all about protecting you and your family.

MODERATOR:

So, what role do you see for the companies in this kind of like governance conversation and framework?

MURPHY:

Well you know, as was referenced, I don't think that there is any system that works without a mandatory system of regulation and approval of models. And I think, ultimately, of integration into different industries. I think we will ultimately have a regime that says: ‘listen, we're going to approve products that are sold to teachers, we are going to approve products that are put into our hospitals to make sure that they are benefiting the public rather than benefiting the companies.’

But obviously this would be better done with the companies sitting at the table. I think certainly at least Anthropic sees the fact that, without a level playing field of regulation, there's no way for one company to deliberately slow down investments in innovation without losing their company because the profit motivation is so extreme that the investors – and all those companies have basically the same set of investors – are just not going to permit a unilateral decision to not move as fast as possible for commercial deployment.

But we shouldn't wait for their permission. We need to do this in the next 12 months. And so, while I would like them at the table, the last thing we should do is give them veto power. They will attempt to try to have veto power. They will try to spend money in elections to get folks in power who will only pass regulatory regimes that they approve. But I think that would be one of the most dangerous approaches we could take: to only pass regulatory regimes that get the sign-off and approval of these companies. That, by definition, might make it illegitimate in the public's mind, that is already convinced that these companies are not out to protect them, but out to protect their bottom line.