Trump Is Burning Down Everything That Kept Us Safe

Stacey Abrams
1 follower
2 Video Views·Oct 5, 2026
AI Generated·May contain errors
Video Summary
AI GeneratedStacey Abrams and Susan Rice discuss the fragility of democracy and global stability, contrasting Brazil's accountability for Jair Bolsonaro with U.S. shortcomings. They analyze the strategic importance of USAID and the UN, while warning that unregulated AI poses existential national security risks, necessitating a global development freeze and stronger civic vigilance to protect democratic institutions.
Welcome to Assembly Required with Stacey Abrams from Cricut Media. I'm your host, Stacey Abrams.
On October 4th, 156 million Brazilians can vote for president. One of the men on the ballot will
be Flavio Bolsonaro, a 45-year-old senator whose last name may sound familiar. His father, Jair
Bolsonaro, is not on the ballot. Why? Because the senior Bolsonaro is sitting in a cell at federal
police headquarters in Brasilia, serving a 27-year sentence for trying to overturn the last election.
The younger Bolsonaro has been clear about why he's running for president.
He told the Washington Post that if he wins, he will pardon his father. And he might win.
Last week, two polls of a likely runoff between Flavio and President Lula da Silva
came back as a tie. Back in August, Lula led by five. If nobody clears 50% on Sunday,
the two of them meet again on October 25th. Former President Bolsonaro is in prison for his
mimicry of Donald Trump's January 6th insurrection. Now, the U.S. should be watching closely to see
if Trump's authoritarian allies intend to learn something from Bolsonaro's comeuppance.
But I want to talk about Brazil today because it achieved something America couldn't.
And now it's facing the questions we're going to face too. Here's what Brazil did. After Jair
Bolsonaro lost in 2022, investigators found a plan to keep him in power anyway. Sound familiar?
According to the charges, it included plots to assassinate the president-elect, the vice
president-elect, and a Supreme Court justice. On January 8th, 2023, two years and two days after
our own January 6th insurrection, Bolsonaro supporters stormed Brazil's Congress, its Supreme
Court, and its presidential palace. But here's the difference. Brazil investigated. Using their
system of justice, the people at the top of the scheme were charged, including the former president
himself. Prosecutors brought evidence, and in September of last year, that court convicted Jair
Bolsonaro of leading an armed criminal organization and trying to violently overthrow democratic rule.
Last November, he reported to prison. But then there's what America did. We convened a bipartisan
committee, hundreds of hours of testimony, and a federal indictment. Then the man at the center of
our scheme won again. And on his first day back in office, he pardoned or commuted the sentences
of roughly 1,500 people charged in the attack on our capital. The prosecution stopped. The case
against him went away. And so for the last few years, Brazil has been the country that actually
got it right. But what's happening now shows that getting it right once doesn't settle anything.
I need us to understand that authoritarianism isn't solved by a single election or a righteous
prosecution, although both are important. Because the perpetrators are crafty and persistent.
They're not idealistic, and they're not deterred. So look at what the Bolsonaro movement did after
his conviction. It didn't fall apart. It went to work on three fronts. First, it went after the
In December, Brazil's Congress passed what they call the dosimetry bill. It rewrites how
coup-related sentences add up. Lawyers estimate that it could cut President Bolsonaro's time
in a closed prison to two years and four months. Now, President Lula vetoed the bill,
but Congress overrode the veto in April. It compromised Congress. In May, the Supreme Court
suspended the law until the full court can rule on whether it's constitutional. So that fight
isn't over. Second, the regime found an heir. The president can't run, so the son is running
on the promise of setting him free. And third, the regime looked for help from outside the country,
and they found allies right here in the highest levels of American government.
In July of last year, Donald Trump sent Brazil a letter calling Bolsonaro's trial
a witch hunt and threatened a 50% tariff. By the end of that month, the tariff was in place.
The U.S. State Department revoked visas for eight of Brazil's 11 Supreme Court justices.
That's Marco Rubio's domain. And then this summer, the U.S. Trade Representative
announced another 25% tariff on Brazilian goods effective July 22nd. Among the stated reasons
was something called Brazil's anti-corruption interference. Think about that. Anti-corruption
interference. Well, Vice President J.D. Vance's tirade against corruption and fraud
apparently stops at the border of democracy. But let's think about what all of this means.
The government of the United States used trade policy, the thing that sets the price of whether
your family can buy something at the store. They used it to punish another democracy for putting
a coup plotter on trial. The perpetrators of this ignominy are Marco Rubio and J.D. Vance,
the likely heirs or inheritors of the MAGA throne. And America's might has been used to foster
Brazilian corruption. For those who continue to see politics as usual operating in the United
States, despite everything we've seen over the last few years, I need us to consider the naked
reality that America's leaders are trying to free a convicted criminal because they prefer
corruption to accountability. And we see this dedication to corruption at another level of
Republican rule. Over in Brazil, at the ruling party's convention in July, Flavio's campaign
played an AI-generated video of his imprisoned father endorsing him. But Jair Bolsonaro never
actually said those words. The video was made by a computer. Brazil's electoral court took the case
and had to decide five weeks before the vote what counts as a deepfake in a campaign. You see,
Brazil has written rules banning deepfakes in campaign ads that are meant to help or hurt a
candidate by imposing penalties up to and including throwing said candidate off the ballot. So whatever
you think of where they drew the line, at least Brazil drew one. Here in the U.S., we still don't
have a federal law on deepfakes or any other kind of AI harms. Instead, we have a patchwork of state
laws. And in most races, nothing stops a campaign from putting words in someone else's mouth.
And lest we forget, a few weeks ago, these same AI giants asked for permission to collude to avoid
even more oversight. As we fast approach the midterms and as we prep for 2028, I want us to
understand what Brazil is screaming for us to learn. First, winning an election is not a protection
against the erosion of democracy. Authoritarians love elections. Competitive authoritarianism
is driven by the hijacking of democratic institutions like elections to give legitimacy
to their corruption and their overreach. We're watching it in America in real time,
and Bolsonaro wants a rematch in Brazil. Two, losing an election does not give permission
to try to overthrow democracy. Look, I know what it's like to lose an election I believed was
unfair. In 2018, I lost a race for governor of Georgia, and I said then and still say that voters
were kept from the ballot. But I also said that my opponent won under the law, and I took my fight to
the courts, the legislature, and to organizing. I did not raise a mob. The line between contesting
an election and overturning one is the whole difference between a democracy and whatever
comes after it. Brazil enforced that line. America didn't. And both countries are now contesting
what comes next. Our midterms are November 3rd, five weeks after Brazil votes. But 2028 is fast
approaching, and the presidential contest of 2028 starts on November 4th. We have too much at stake
to misunderstand the moment and the lessons to be learned. So here's what I'm asking you to do from
our 10 steps to freedom and power. First, I need us to commit. Learn the pattern Brazil just showed us.
Winning the election is not enough. After the midterms, regardless of the outcome, we have to
watch what happens next. We cannot be lulled into believing that a single election will stop the
march of white Christian nationalism, proto-natalism to put women in their place, or xenophobia to
stoke fears of immigrants. They will simply put new faces on their intentions, same names, same ideas,
and they're coming to a state legislature near you. So commit to paying attention. Next, we've got to
mobilize. Be at the polls this November, but also be in community. One of the Bolsonaro tactics was
to use outside help from the U.S. to undermine internal political consequences. Don't let that
work here. So whether it's snap cuts that will hit families right after the midterms, or false
accusations of fraud that are being used to deny access to health care, we must show up in
congressional offices and at state legislative hearings to demand accountability and good policy.
Otherwise, they're going to start to blame Democrats for Republican policies, and they're
going to blame vulnerable people for billionaire tax breaks. Third, elect. Early voting is already
open in some states. Voter suppression is on the rise across the country, from poll closures in
Tarrant County, Texas, to laws banning water here in Georgia. Reach out to five of your neighbors
and ask if they have a plan to vote. That's how we fight back. And then, once they've told you,
ask how you can help make sure that plan happens. The best protection for an election that is under
threat is too many voters, people who believe they have the right to be heard, who show up in droves
and demand that we listen. Brazil held its line once, and it's being tested again on Sunday.
Our test is in five weeks, and there's still time for us to get ready. Joining me this week is Susan
Rice. She's held many roles, including as domestic policy advisor to President Biden,
as a national security advisor for President Obama, and as the U.S. ambassador to the United
Nations. Simply Required with Stacey Abrams is brought to you by Smalls. Like so many of you,
I have some friends who absolutely love their cats and treat them like family. So, of course,
they want to give them the best food possible. And having served as a host for my sister's beloved
cat, Pepper, I personally know that cats can be surprisingly particular about what they'll eat.
That's one reason Smalls caught my attention. Smalls makes fresh cat food with 100%
human-grade ingredients, packed with protein and gently cooked without preservatives,
and delivers it right to your door. The results are pretty impressive. Smalls says that 88%
of cat owners reported seeing overall health improvements after switching their cats to
Smalls. That can mean things like better digestion, more energy, a softer and shinier coat,
and fewer hairballs. The food is designed around what cats actually need. Smalls is high in
protein and helps with hydration, giving cats more of the nutrients they need to support strong
muscles and sustained energy. And if you're a little nervous about committing to a new food
for your cat, you can try Smalls risk-free. They'll refund you if your cat won't eat it.
So stop serving your little carnivore a bowl of processed shortcuts. For a limited time,
because you're an Assembly Required listener, get 60% off your first order, plus free shipping
and free treats for life, when you head to Smalls.com slash Assembly. One last time,
that's 60% off your first order, plus free shipping and free treats for life, when you
head to Smalls.com slash Assembly. Assembly Required with Stacey Abrams is brought to you
by Honeylove. I thought I couldn't do without underwires, and then I tried Honeylove. Their
crossover contour bra is a great example of why wires aren't always necessary. It combines their
best-selling wireless design with molded foam cups to create a smooth, natural shape. What's
CloudFuse bonding? It's a miracle of engineering to create a terrific bra. There are no wires
involved, so there's no poking, digging, or pressure points. I've tried other no-wire options,
but they rarely worked out. Honeylove got it right. Why? Because Honeylove is an independent,
female-founded brand. All Honeylove products are intelligently designed by women who actually wear
them, including founder Betsy, who's highly involved in the development of each product.
Their bras are designed to provide lift and structure, using innovative bonding technology
instead of uncomfortable underwire. So you get that secure, lifted feel without the poking,
pinching, or constant adjusting. And their Cloud Embrace bra is a game-changer. Soft,
breathable, and incredibly comfortable, but still supportive enough to make you rethink your
underwire bras. Because comfort isn't a luxury, it's a requirement, and support shouldn't mean
sacrificing how you feel in your own skin. And it doesn't stop at bras. Honeylove shapewear is
especially worth calling out. It's designed to smooth and support without squeezing or restricting,
using targeted compression that works with your body. Whether I'm home on calls all day or on a
speaking tour, this is a product that moves with me and keeps me comfortable all day long.
Honeylove products use high-quality fabrics that feel great against your skin. Thoughtful
touches like breathable materials make these pieces as practical as they are comfortable.
At the end of the day, this isn't about changing your body. It's about supporting it in a way that
actually feels good. And Honeylove delivers that at every stage. Treat yourself to the most advanced
bras and shapewear on the market. Use our exclusive link to save 20% off Honeylove at
honeylove.com slash assembly. That's honeylove.com slash assembly. After you check out, they'll ask
you where you heard about them. Please support our show and tell them we sent you. Experience
the new standard in comfort and support with Honeylove. Susan Rice, welcome to Assembly
Required. Great to be with you, Stacey. So we've had a busy two plus years, but
I'm going to start with last week and the U.N. We saw a lot of news being made at the U.N. and I
want to start with your role as a former U.N. ambassador and just the big picture. Part of what
I think people could take away from last week's shows and side shows and shenanigans is a question
about the continued utility of the United Nations. Can you talk a little bit about how you see the
role of the U.N. in a moment where we are seeing such a fracture in international accord and this
continued assault on democracy across the world? Well, you know, the U.N., as folks will recall, was
created right after World War II by the United States as the principal architect to try to prevent
World War from recurring and to prevent and resolve conflicts and to uphold human rights and
the sovereignty of nations. And, you know, today we have a president of the United States who goes
to the U.N. and stands before the General Assembly at that very, you know, hallowed podium and very
casually threatens to annihilate another U.N. member state as problematic and dangerous as that
government may be in Iran. And so we really have come full circle to a time when the United States
at least purported to stand for stability, international law, security,
and human rights to a time where we, you know, joy in the trampling of all of the above.
And it does make it extremely hard in that context for an organization like the United Nations,
very flawed as it is, to even serve a fraction of the purpose it's meant to. Now, the U.N. has got
many different components and most people think about, you know, the Security Council where the
big five permanent powers, U.S., China, Russia, the United Kingdom and France, you know,
make decisions about the issues of the day. A part of the U.N. that many people don't think
about and what I would argue is its most important element right now in the current context are the
operational agencies of the U.N., the humanitarian agencies, things like the World Food Program,
the High Commission for Refugees, UNICEF, the World Health Organization, which are doing the
work on a daily basis around the world to try to contain Ebola or feed starving children in Sudan
or the peacekeepers who are on the front lines of conflicts that most people have forgotten about.
Those are the elements of the U.N. that I think are most relevant. Sadly, the Security Council,
where I was privileged to serve for four and a half years in what seems like a long, long time
ago, is paralyzed by dysfunction on the most important issues from Gaza to Ukraine to Iran
because the United States, China and Russia are almost all the time at odds on the issues that
matter most and each wields a veto and so nothing can be accomplished unless all of us agree.
I want to go back to something you just said about the functional responsibilities of the
U.N. to solve problems, many of which have been exacerbated by American withdrawal,
American undermining or American aggression. And in particular, I want to talk a bit about
the fallout from the dismantling of USAID. And I know this has been something you've
been thinking about and talking about for a while, not only from your vantage point of
having been a U.S. ambassador to the U.N., but also as someone who sat as a domestic advisor
and who sat as a national security advisor. And I think there has been some misinformation about
why USAID was so critical. Can you talk about why USAID was such an important part of national
security and what it means when we pull out not only our own money, but when we have a dysfunctional
approach to the U.N.? Well, dismantling USAID was one of the most serious own goals that the
United States has committed in my lifetime. USAID was set up in the 1960s with the aim of making
Americans and the world safer. Now you ask maybe how does providing assistance to foreign countries
or foreign nongovernmental organizations make the United States safer? Well, the work that USAID
did was about preventing and resolving conflicts, dealing with refugee flows, hunger, the fallout
from climate change, preventing diseases from spreading when we could help to detect them early
and contain them, preventing countries from going to conflict over things like water because, you
know, we were able to provide life-saving humanitarian assistance, food, water, medicine,
and so much more through USAID. Helping countries to become economically viable, stand on their own
two feet and become trading partners for the United States, export markets for the United States.
That's what USAID did on a daily basis and, you know, it became, you know, fun and games within
some quarters of the Republican Party to denigrate USAID. But really that was the fringe, you know,
most of USAID's history has been, had been characterized by bipartisan support for the
work of that organization because Republicans and Democrats alike understood its value, its
centrality, in fact, to our national security, which is why you've heard so many four-star
generals say, you know, if I don't have USAID, I'm going to need more weapons and bombs because
they understood that the work of USAID, its development assistance, its humanitarian
assistance, its promotion of democracy, were things that made Americans safer and actually
served to prevent conflict. So I sit on the board of the National Democratic Institute, which has a
Republican counterpart, the International Republican Institute. I used to be on that same board once,
back in the day. I remember. And I'd love for you to talk about, just for a moment, because you
raised the issue of bipartisanship. And when we saw this decimation of investment of American
capital in foreign aid, we also saw this very concentrated and sustained attack on the NED,
which is the National Endowment for Democracy, which helped support both NDI and IRI. And it's
one of the more quietly successful pushbacks that we've been able to maintain some degree of
bipartisanship on the issue of investment in democracy. But we also know that it continues
to be under assault. And sort of pulling that thread that you just laid out, why is it a bipartisan
issue to maintain operations for democracy around the world? Why should that be something that
creates common cause instead of being weaponized right now? It's because the United States is safer
when there are more countries and peoples who share our values and share respect for human
rights and believe in and implement democracy. Democracies are far less likely to go to war with
one another. Democracies provide for the aspirations of their people, at least in principle.
Democracies are America's allies and friends historically. And we're safer in a world where
we have more friends and we have fewer. Now, that may sound like a radical concept in the current
context because you see in this administration a complete disregard for friends and allies,
an effort to undermine and weaken them. Look what we're doing to our closest friend in the world,
Canada, and neighbor, for one, Europe and Asian allies similarly. So this is an administration
that I think sees the world through an entirely different prism than all of its predecessors.
America has benefited enormously in terms of our security and our geostrategic position,
as well as our economic well-being, by having allies and friends around the world. Our alliance
network is the envy of our adversaries. And yet, now we have a president who is
doing just what our adversaries have long dreamed, which is to undermine and work to
ensure the fragility of those alliances. So democracy promotion is part of how we broaden
the scope of peoples and countries that are inclined to partner with us when we are
behaving in a manner that's more consistent with our values.
When we come back, more from my conversation with Ambassador Susan Rice.
Assembly Required with Stacey Abrams is brought to you by BILT.
Here's a question. Are you getting rewarded for paying your rent or mortgage right now?
If the answer is no, you need to check out BILT. Becoming a BILT member means earning
points on every housing payment, no matter where you live. BILT points can be redeemed for flights,
hotels, Lyft rides, Amazon.com purchases, future housing payments, and more. Personally,
I'd use them for BILT home delivery powered by GoPuff and at select restaurants.
An unexpected benefit BILT gives every member access to is its agentic Neighborhood Concierge
in the BILT app. It doesn't just give you information about your home and neighborhood,
it executes. It's connected to your home and over 50,000 merchant partners, so you can ask it to
pay your rent, make a dinner reservation, book a fitness class, schedule a Lyft, and more.
It makes things happen while automatically applying your member benefits along the way.
For a crowd that thinks a lot about what agentic AI actually looks like in practice,
this is an example worth paying attention to. So first, download the BILT app. Second,
link any credit card or debit card to the app so purchases at your local spots become even
more rewarding. And third, ask the Neighborhood Concierge in the app pretty much anything.
It'll tell you about exclusive member benefits at restaurants, fitness studios,
on your Lyft rides, and more around your home. And for the homeowners out there,
BILT works for mortgages too. Your housing payment is one of your biggest monthly expenses,
so why wouldn't you want to earn points on it? Download the BILT app and join the membership
for where you live at joinbilt.com slash assembly. That's J-O-I-N-B-I-L-T.com slash
assembly. And use our link so they know we sent you. Assembly Required with Stacey Abrams is
brought to you by Forkful. Eating well can be difficult when life gets busy. That's where
Forkful comes in. Chef-cooked, fresh, never frozen, ready-to-eat meals made with high-quality,
premium proteins, grass-fed bison, wild-caught salmon, and fresh vegetables. And they have
delicious menus that fit your lifestyle and meal preferences, including gluten-free, dairy-free,
keto, plant-based, GLP-1-friendly, and low-carb meals. Eating well doesn't have to be complicated
with fully prepared, deliciously healthy meals. No meal prep, cooking, or cleanup.
Just great flavors and ready in under three minutes. Plus, they're delivered right to your
door within 24 to 48 hours. When the days get crammed with meetings and events or a book
deadline, I like that I can pop a great meal into the microwave and start eating in minutes,
like their Peruvian chicken with aji verde. Explore their plans and menus of fresh, ready-to-eat
meals at forkfullmeals.com and use code ASSEMBLY to get 50% off your first order.
That's forkfullmeals.com slash ASSEMBLY, and use promo code ASSEMBLY for 50% off.
And thank you to Forkful for sponsoring this episode. One of the challenges with being
consistent in our values is how we get our information, where information comes from,
how it's disseminated to us. And in the middle of that right now, we have an issue with AI,
with artificial intelligence. And I want to spend a few minutes. Recently, you wrote,
I think, a very thoughtful piece in the New York Times about how we should regulate AI.
And I want to frame out why I think that's a conversation that dovetails with national
security. There's the who's creating AI, that's what AI is actually able to generate.
And then there's the question of what do we think we know about AI? And I want to start with the
national security piece for a second, because Bloomberg recently had a piece about the failures
that have been embedded in military decisions, including the US airstrike on the Iranian
elementary school at the start of this war, in part because they relied on AI for targeting
processes, because the administration had fired 90% of the people who were supposed to be
preventing civilian casualties. So we were using AI to solve a problem and actually kill people.
CNN had a piece about how AI nearly started a war with China, because it misunderstood what
a ship was carrying. And you talk about the necessity of regulating AI. So I'd love to
start that conversation with you from the vantage point of what does it mean as a national security
issue to have a regulatory scheme for AI? Well, let me deal with the kind of incidents
you just described and then take it to a slightly broader level. Using AI
to provide decision makers with information that might be faster and
better than what they might otherwise have, is not inherently the problem. The
problem is when you take humans out of the loop and you substitute human judgment for AI.
So the examples you gave when we almost apparently hit a Chinese ship because we had
bad AI, thankfully somebody looked at that, a human being looked at it and said, wait a
minute, this doesn't add up. When we use AI for targeting decisions for the lethal use of force,
and we don't have humans second guessing, double checking and overriding AI, we do the sort of
thing that we saw with what appears to be the strike that we executed in Iran against that
girl's school. And then imagine that in the context of far more powerful weapons,
including nuclear weapons. So on a fundamental level, we absolutely have to ensure that humans
are the ones making decisions about the lethal use of force. And that should seem like
motherhood and apple pie. And yet, there are countries and there are governments,
potentially it seems including our own, that object to that. The whole anthropic blow up
where the government designated anthropic basically an enemy of the state, a supply chain
threat, was because anthropic was saying, you can't use our large language models for
targeting unless a human is in the loop. And you can't use it to surveil American citizens.
And that was something the government was unwilling to accept. So that's one problem
as it relates to national security. But the argument I was making in that piece in the New
York Times was there are two other elements of national security that in some respect,
I would argue, are even more important. There are two fundamental problems with the
safety of AI models that directly affect national security. The first is that we now know,
and we've known for quite a while, that these more sophisticated models can be used by bad actors,
whether it's a terrorist organization, a hostile nation state, some individual sitting on his bed
in his basement, for lethal attacks in the real world. So we've seen it already with cyber.
You can create highly sophisticated cyber weapons that could potentially knock out whole
electricity grids or take down the internet. You could use AI, particularly the most sophisticated
models, to learn how to make deadly weapons of mass destruction. So therefore, AI has the
potential to be used in a very dangerous way, for example, potentially to create a new super
pathogen that's highly transmissible, highly lethal, that far exceeds the most dangerous
ones out there today. So that concern is very, very pressing. And it's already with us, in a
sense. Anthropic recently reported that it detected its models being used by what they think was a
hostile nation state for biological weapons innovation. The other national security concern
about AI is the risk, and we're seeing this now manifested, that humans could lose the ability to
control the agents. And the agents can go out and do nefarious things on their own, in essence. And
we've seen swarms of agents escape training runs inside of, you know, get out of their sandbox,
get out onto the internet, and start hacking into other companies, governments, whatever.
A loss of human control over artificial intelligence is extremely concerning,
and that's where you get to those saying that this could result in human extinction.
Now, you know, is that a probability or just a possibility? I don't know that you or I could say
that with certainty. The people who know the technology the best say it's a decent possibility.
And so to me, that's completely unacceptable. And we have leaders, particularly in the White House,
who are whistling past the graveyard, aware of these risks and unwilling to do anything
meaningful to stop them. And we have the heads of these companies saying, well,
I love the Saturday Night Live takeoff of Dario Amadei, who is actually one of the more thoughtful
and responsible of these leaders, but basically saying, help me, you know, I can't stop myself.
I know I'm wrong, but I can't stop myself. Well, that's also ridiculous. And the argument I was
making, Stacey, and I'd love to get into it in more depth with you, is that in a rational world,
which I grant we may not be living in right now, you know, it ought to be possible for the United
States and China to collaborate to solve the two most serious problems, the problem of misuse by
terrorists or non-state actors and the problem of loss of control, which the Chinese have an
extraordinary interest in. The last thing the Communist Party of China wants is to lose control
of anything. And that is an interest in which, you know, as humanity, we have a shared stake. And yet,
you know, Trump squandered the summit last week to deal with this in a meaningful way. He says
it's a hoax. He doesn't want to deal with it. Yes, we're in a competition with China, but there
is a way for us to approach this with the Chinese that would make us all safer and not affect the
competitive dynamic over the long term. And I'd love to hear your thoughts and I'd love to elaborate
on that. Absolutely. And I'm going to do a quick commercial for myself, but only because I think
it's helpful. I'm going to do it for you. I know where you're going and you stole it from me.
You wrote a great book on the threat of A.I., a novel, one of your wonderful Avery Keene series
called Coded Justice. And I told you after you wrote it that I loved it. And you deal with the
real challenges of A.I., particularly on the misuse. So I'll let you go from there. But I
just want to say I love that book and everybody ought to read it. Well, thank you. OK, that's a
high compliment. And it was not a setup. I was actually going to mention. No, no, because I
mentioned that one. And then Rogue Justice, because one thing you just said when you talked about the
cyber attacks on the energy grid. So Rogue Justice, which was the novel before Coded Justice,
and they're all standalones, you can just go and pick them up. It looks at what happens when you
have concentrated attacks on our energy grid and how little we understand about that. And this
comes back to what you've said so eloquently in that piece, which is we don't know what we don't
know. And the people who are in charge of telling us are telling us they also don't know. And a lot
of the work that you've led, both on the domestic side and on the national security side, has been
about confronting our blind spots and our ignorance. And one of the challenges I have,
I was recently at an AI conference and had a very long and thoughtful conversation with someone
who works for one of the big three. And he basically said we should not be allowed to
touch AI. We need to stop everything. And so I would love to have you talk a little bit more
in depth about why the possibility of just freezing development is not something we should
reject out of hand, to start there. So if you're about to drive off a cliff,
the only rational thing to do is to slam on the brakes. And that's the situation we're in.
We are about to drive off, we being humanity, are about to drive off a cliff and risk,
we will certainly crash the car and we risk killing everybody in the car. So how do we
put the brakes on? Let me share what I suggested in this piece I recently wrote,
but then come back to, well, the obvious question, well, what if that doesn't work,
which is more likely than not? The problem that we have, as folks know, is that AI is evolving so
quickly that it is reaching the place where it is going to be able to improve upon itself. It's
called recursive self-improvement or RSI. Basically, it is not only self-perpetuating,
but self-evolutionary. It is making itself by itself more and more sophisticated. And that's
how it exceeds human intelligence potentially and becomes artificial superintelligence.
At which point, we are not in control. And so that's obviously not a place I want to go. It's
not a place I want for my children to go. And so what do we do? Well, the labs all say, if we slow
down, then our competitors will get ahead, whether they're talking about their domestic competitors
or more often they try to say, well, they're Chinese competitors. China will win the AI race,
and that'll be basically the end of our way of life as we know it. I'd like the United States
to win the AI race, but only if we can win it safely. And when I say safely, I mean for humanity.
And where we are now is we're just racing all of us off a cliff as fast as we can. My
thought was, my proposal is, that the President of the United States and the President of China
order their respective AI companies to freeze in place their advanced development and training and
deployment of their AI models. Not only those that are in the frontier labs that are like
open AI and anthropic, but also those that are called open source. These are in some ways
both more attractive to many because they're cheaper, they're more accessible, but they're
also more dangerous because they come with guardrails that anybody, any user can strip away
in order to customize it for their own purposes. The Chinese are excelling in the open source
models in particular. So it's got to be both these types of models, the frontier models,
open source models that are the most sophisticated. Freeze in place their development until
a team of the world's best AI scientists and experts and academics can be assembled. And I
think the US and China could pull such a team together, not only of their Americans and Chinese,
but the best experts around the world. Give that team the task of solving on a technical,
scientific basis, these two big challenges of misuse of AI for weapons of mass destruction
or cyber attacks and loss of human control. Because right now, these guys are racing to
get to superintelligence without regard for the safety imperatives or without sufficient regard
for the safety imperatives. Nobody has solved the problem of what they call misalignment. How
do you prevent these models from doing stuff we don't intend them to do? And yet we keep
rushing toward the cliff. Let's stop before we get to the end of the cliff. And let's say,
wait a minute, let's put some high quality brakes on this car so that it will stop when we hit the
brakes. And that's an engineering challenge. That's a technical challenge. If this world-class
all-star team of scientists, all on the task of solving these two problems, cannot solve the
problem, then we shouldn't keep driving the car. We should all say we're not going to
superintelligence. But assuming that with concerted effort, they solve the problem,
then that solution, that technical solution would be required to be embedded in every
subsequent model ever released. So that doesn't solve all the problems of AI. We're still going
to have issues of disinformation and how it messes up our kids and how it screws over our
economy. And will the US and China compete in the military realm in ways that are destabilizing?
All of those are really important problems we have to address. And we need legislation to do so.
But this problem of misuse and misalignment has to be the top priority of the best minds in the
world. They created this damn problem, let them fix it. And if they can't fix it, then to hell
with them. We shouldn't be living in a world where anybody can drive us off a cliff.
Well, I'm going to extend your analogy a little bit because I think there are those who hear the,
and we saw this with Jensen Wang, you know, oh, this is overblown. The, you know, RSI problem is
not a real problem. And then he hedges his bet and says, well, and if it is a problem,
we can solve it ourselves, which I think ignores one piece of the analogy, which is that, yeah,
we're heading for a cliff, but we're hitting a lot of things along the way. There's collateral
damage all on the road. And we keep thinking about it. And the moment the car goes over the cliff,
we forget about all of the things that happen on that journey. And what is happening when
we have the hugging face incident? One of the things I speculate about in Coded Justice is
what if the intentions aren't evil? What if the intentions are good? Because good intentions do
not forestall bad outcomes, Oppenheimer, anybody. And so one of the things I'd love for you to talk
about for a second is, you know, what do we do about the things that are happening along the
way? And what do we do about the people who aren't included in the conversation? Yes, China and the
US are the leaders, but France is doing a pretty good job. Switzerland is building an open public
AI model. There are nation states that can use the technology that's already been built. And this
brings us full circle back to the UN, where we don't have an international conversation happening,
because we're treating it like a two person arms race, when we've got some open source,
you know, plutonium out there and uranium out there. Yeah, agreed 100%. And in my proposal,
it's already long and complicated enough. So thank you for giving me a chance to plug some holes.
So in my conception, the US and China would reach an agreement to call to demand that their
labs stand down and freeze in place. But they would also take it upon themselves to prevail on
the leadership of other countries to do the same. So that all of the countries, and you mentioned
France, and there's the UAE, and there are others that are, you know, they're doing pretty well,
they may not be quite where the US and China are. They have to stop too. And they have to
participate in this freeze. And, you know, I also say that during the pendency of the freeze,
we ought to do two things in parallel. One is get our act together finally and pass some
meaningful domestic legislation that deals not only with the safety and security concerns,
but deals with the other things we touched upon, the economic and employment implications that
deal with children and mental health, and, you know, disinformation and misinformation,
and the whole panoply of things that we have not touched with a 10 foot pole. Meanwhile,
the Chinese have regulated for those things. And then secondly, to codify an understanding
between the United States and China that deals with what are the appropriate uses of AI,
and then to go together to the UN to the international community and try to
build that out into a binding international treaty. It's not dissimilar, Stacey,
to the way the United States and China collaborated, came to an understanding
on climate issues in advance of the Paris Climate Agreement. And then, you know,
that got broadened into a broader international understanding. So I'm not trying to suggest the
U.S. and China are the only countries in the world that matter by no stretch, but when the U.S.
and China, who are in the lead on these things, can come to an agreement, as we've seen on this
and other issues, it is possible then to broaden it to the rest of the international community.
Now, some people might say, well, you know, how are we ever going to agree with China? You know,
there are competitors, they have no interest in doing this. You know, there is no guarantee.
Obviously, I called this proposal a moonshot, because it is, it's a diplomatic and technical
moonshot. But the Chinese have some shared interests with us. They don't want terrorists
getting a hold of biological weapons any more than we do. And they don't want AI going rogue
and eluding the Communist Party's control. That's existential for them. So presumably,
theoretically, and I think actually, we have some common interests, even if we're then going to go
back and compete on other dimensions. The United States, then you'll say, well, look at Trump,
he's calling it a hoax. He's saying full steam ahead, seems to have no regard for the condition,
the concerns we've just been talking about. And you know, that is, is big or bigger a challenge
than getting the Chinese on board. But it's interesting within the administration,
there are differences of opinion. And, you know, that Trump is vacillated wildly from,
on the one hand saying, you know, let AI rip, do whatever it wants, full steam ahead.
And the other hand, he took a bazooka to anthropic and nearly made a decision that would have put it
out of business back in June, which when it was the leading company for the United States.
So you don't know where he's going to land. He can flip flop, as we've seen on Iran, on,
you know, tariffs. And so maybe, you know, who knows what happened at dinner between him and
Dario, maybe Dario got through to him on some level. You know, he seems to go with the last
person who talked to him. So I don't think it's impossible, even if it's a moonshot.
So let's think about what we do just in case Dario is not magic.
So on Assembly Required, we believe in giving our listeners homework. So we got,
we got a few thousand folks out there who are going to listen to this and who are going to be
a little terrified, but then think, what can I do about it? And we've spent time on the show
really trying to unpack AI and explain its domestic and international implications. And I
think what was so important to me about your essay was that it actually gave a framework and said,
here are some things we can do. But let's bring it down to the person who's listening to this
podcast thinking, what can I do? What homework can you give them, whether it's related to elections
or related to the day after elections? What can the average person do to make the moonshot possible?
So let me answer that. But let me also say that, you know, in a world where we don't come
to an agreement with China or we don't come to an agreement with China quickly enough,
we do need to put some very serious brakes on what we're doing here domestically. We should
be banning recursive self-improvement until we know that it works. We should be ensuring that
the labs are not able unilaterally to drive us over a cliff. To me, as much as I'm a national
security person and believe that we want to be in the lead over China on this, as I said earlier,
not at the expense of Americans, not at the expense of humanity. So I will say that.
Now, what do we need? What's our homework? The first piece of homework I was going to
assign was coded justice, but that's more for fun. But the most important thing is for the
American people to understand what's at stake here and not let, you know, a handful of people
in the White House and in Silicon Valley dictate our fate and our future. And the way to do that
is by exercising our voice and our vote. Critically important that we put pressure
on our lawmakers in Congress, state and local level, and on the White House to address this
problem. I'm really, you know, I get and I'm encouraged by the focus that people have recently
put around data centers, but that is kind of the symptom that touches people's neighborhoods and
daily lives that we got to get at the root cause. And so we need, in my opinion, to make our
decisions at the ballot box in part based on who's taking this issue seriously and who's prepared
to put the brakes that we so badly need on this. So like I imagine many of your shows and
Stacey, you know, voting is critical. Using your voice and your vote is absolutely essential. And
I would just add one other thing. In addition to getting out and voting and bringing, you know,
as many friends and family as you can with you to the polls, I'd encourage folks also,
particularly younger folks, to get involved by being poll workers and indeed poll watchers.
We have, you know, the poll workers who are out on the front lines every year at your library or
your elementary school. You might notice they're kind of old and they're well-seasoned, right?
And, you know, they're also coming under a lot of pressure. Some of them are being harassed and
intimidated. This ought to be a young person's job. And we, as a matter of our political culture,
ought to get into the habit of, you know, encouraging people of all ages, but in particular,
younger people, to take on this civic duty and to do it, you know, not just once, but every cycle.
So being poll workers, being poll watchers, helping to be on the front lines to ensure the
integrity and the trust in our elections, in addition to doing the work of educating yourself,
using your voice and ensuring your vote is cast and counted.
Susan Rice, Ambassador, Advisor, and Potentate, thank you so much for being here today on
Assembly Required.
Great to be with you, Stacey. Thanks so much.
Thank you.
As always at Assembly Required, we give you actionable tools to help make a difference.
First, be curious. To learn more about the companies that, for better or worse,
are shaping our AI future, pick up a copy of Karen Howe's excellent book,
Empire of AI, Dreams and Nightmares, and Sam Altman's Open AI.
Number two, let's solve some problems.
The 10 Steps Campaign continues to add new partners.
So please visit 10stepscampaign.org partners to find a local cause you care about.
Today, I want to spotlight the Southern Legal Center for Youth,
which provides free civil legal support to young people ages 12 to 24 in Georgia,
especially those who are homeless or at risk of experiencing homelessness
without parents or guardians and LGBTQ plus youth needing gender-affirming legal care.
Visit slcy.org to learn more, donate, and get involved.
They are always looking for people who can sponsor legal clinics
and attorneys and law students who can volunteer.
So show up.
And lastly, let's do some good.
The East Coast has been pummeled by storms, which can cause flooding and power outages.
So check on your friends and neighbors,
especially those who are elderly or in need of medical support, and make sure they're saved.
Thank you to those of you who share questions and comments via my Substack,
Assembly Notes, Crooked Media's Discord channel, and other platforms.
When you like, subscribe, and share, you feed the algorithm
and help others find their way here.
So do me a huge favor and keep the questions and comments coming.
And if you like what you hear, tell others about us and add us to your feed.
We keep growing.
We're closing in on 100,000 subscribers on YouTube with your help.
So however you listen or watch, make sure you show your support and bring your friends.
And then let me know what episodes resonate and what you'd like to learn more about.
That wraps up this episode of Assembly Required with Stacey Abrams.
Do good out there, and I'll meet you here next week.
