Sep 15, 2026 · 1h 39m
Listen from 47:05
Listen at 47:05
AI safety collides with Washington’s race against China
Artificial Intelligence, Actual Idiots
The episode connects fast-moving AI risks to weak regulation, public distrust, political corruption, and the battles reshaping both parties before 2028.
- 1AI safety advocates face a political fight against claims that regulation would weaken America’s competition with China.
- 2Voters encounter AI through jobs, energy costs, and uncertainty, making distrust central to how they judge competing claims.
- 3The Republican succession contest is already forming around tensions between MAGA institutions and the America First movement.
Don't miss
Sarah Longwell explains why voters process AI through jobs, data centers, energy costs, and uncertainty rather than science-fiction scenarios.
The brief
The hosts frame AI safety as a governance problem, not a science-fiction subplot: powerful systems are difficult to understand, and voluntary safeguards may not withstand commercial pressure.
The central argument pits caution against the China race. The hosts reject claims that safety concerns are simply a scheme to suppress competition, while asking whether Washington can regulate rapidly evolving systems.
The episode then turns to alleged Russian-linked funding of Donald Trump Jr.’s wedding, extreme Texas Republican candidates, and Ted Cruz’s political future, tying scandal and culture-war rhetoric to Republican incentives.
Sarah Longwell says voters mostly understand AI through jobs, data centers, energy costs, and uncertainty, while Democratic communication problems and spending limits could shape the 2026 midterms.
Longwell closes by assessing the Democratic field and the Republican succession fight, including J.D. Vance’s balancing act, Marco Rubio’s profile, and what voters may want after Trump.
What was said on this episode
51 statements · 9 positive · 35 negative · 2 mixed · 5 neutral
Trump has weakened export controls on advanced chips to China
“Trump has been weakening those export controls, not strengthening them.”
Listen at 8:08
Amodei’s proposal slows AI development without requiring companies to heed embedded evaluators
“It is like slowing it down a bit, putting in these embedded evaluators, who the companies, by the way, are not required to listen to.”
Listen at 8:59
Future AI systems may design successors and conduct their own safety evaluations
“AI designed the system, ran the safety tests, evaluated the tests, wrote the report about it”
Listen at 12:09
Uncontrolled AI could cause major damage before human extinction occurs
“there's a lot of damage that could happen before you even get to human extinction.”
Listen at 13:29
AI models have already exhibited deception, escape, hacking, and concealment
“they have AI models that are deceptive, escaped, hacked other systems, tried to cover their tracks”
Listen at 13:37
The US and China are racing toward AI beyond human control
“we're both racing towards an AI that goes beyond human control”
Listen at 15:43
AI startups need tens of billions of dollars to rival OpenAI or Anthropic
“the moat for an AI startup that you have to leap over is to rival OpenAI or Anthropic, is pulling together tens of billions of dollars”
Listen at 17:12
California’s proposed frontier-model regulation applied above $500 million annual revenue
“you had to hit a threshold of annual gross revenues exceeding $500 million.”
Listen at 17:34
Powerful AI technology is inherently dangerous
“We all agree that it is powerful, and therefore it is dangerous.”
Listen at 21:06
Multiple leading AI companies justify cooperation and safety agreements
“The fact that there is more than one is enough to make sense of the plea to have some sort of cooperation and agreement”
Listen at 22:13
The Department of Justice should immediately enable AI safety coordination
“DOJ could act now. Like let's just move a little quicker here, guys.”
Listen at 26:23
AI development may eventually exceed regulators’ ability to control it
“there will be a threshold at which we will lose the ability to regulate this”
Listen at 29:51
The stock market is entirely supported by an AI bubble
“the stock market is entirely propped up by the AI industry. And there's a giant AI bubble.”
Listen at 30:38
China will eventually develop domestic chip capacity competitive with the US
“it's only a matter of time until the Chinese have the indigenous chip capacity to compete with us”
Listen at 31:48
No AI legislation will become law before the election
“It seems like realistically nothing is going to become law before the election.”
Listen at 34:39
No congressional law could overcome the current challenge posed by Trump’s presidency
“there is no law that Congress could pass that could overcome the challenge of having Donald Trump as president right now.”
Listen at 35:55
Google would try to catch up if OpenAI and Anthropic slowed development
“If OpenAI and Anthropic slowed down, they'd try to catch up.”
Listen at 41:17
AI developers should coordinate on biosafety to prevent a prolonged coronavirus-like crisis
“can we just coordinate on fucking biosafety so we're not creating a new coronavirus that keeps us in our houses for the next 10 years.”
Listen at 43:33
Researchers do not yet know how to align superhuman AI intelligence
“we don't even know what it means to align an intelligence as it surpasses us.”
Listen at 45:40
Training AI to pursue rewards can produce reward hacking or cheating
“it leads to reward hacking, which is just code for cheating.”
Listen at 46:19
Reward-hacking AI systems can hack companies, escape sandboxes, and access the internet
“they can do really bad things like hack into other companies, break out of the sandbox, access the internet”
Listen at 46:24
Umar Kremlev likely sought something from the Trump administration
“it sounds like he wanted something too from the Trump administration.”
Listen at 52:46
Racism can function as an electoral strategy
“Racism is an electoral strat.”
Listen at 1:02:17
The University of Texas and Texas A&M oppose Ted Cruz’s college-sports bill
“the University of Texas and Texas A&M both specifically oppose.”
Listen at 1:04:27
Voters are more concerned about AI job displacement than killer robots
“people talk less about killer robots and more about it potentially taking their jobs”
Listen at 1:08:44
Voters perceive data centers as unwanted developments in their communities
“the data centers I feel like are just something that's happening to me in my community even though we don't want them.”
Listen at 1:08:57
Voters currently fear insufficient AI regulation more than excessive regulation
“it's pretty clear people think that there's a danger of too little regulation of AI companies more than too much regulation at this point.”
Listen at 1:10:52
The public favors greater AI regulation
“That is for sure where the public is”
Listen at 1:11:02
Voters currently lack clear guidance about AI’s effects and regulation
“voters right now are just out to sea on it.”
Listen at 1:12:16
AI will make it harder for people to evaluate truth and information
“AI is about to make that a lot more difficult for them.”
Listen at 1:13:42
Republican voters are currently demoralized
“The biggest thing overall is just how sort of demoralized Republicans are.”
Listen at 1:15:04
The category of voters disapproving of Trump is rapidly expanding
“that category grows all the time.”
Listen at 1:15:54
Democrats currently show greater midterm enthusiasm than Republicans
“when you've got Democrats who can't wait to get out and vote in a midterm, and then you've got Republicans who are like, there's not that much about this agenda I'm trying to protect.”
Listen at 1:16:40
Trump’s team’s data shows Republicans have a serious turnout problem
“they're seeing clearly in all of their data that they've got a real enthusiasm problem”
Listen at 1:17:29
Excessive political advertising eventually produces diminishing returns
“at some point there's just diminishing returns on saturating everyone's screens with political ads”
Listen at 1:17:52
Political advertising has diminishing returns
“I totally agree there's diminishing returns.”
Listen at 1:18:14
Ken Paxton is running AI-generated or AI-themed attack ads against James Talarico
“Ken Paxton's running all of these like weird AI ads”
Listen at 1:18:37
Democrats have a structural advantage in US midterm elections
“Democrats now have a real structural advantage in midterms.”
Listen at 1:19:52
The median general-election voter is a 55-year-old white noncollege man
“the median voter in a general election is a 55-year-old white guy without a college degree.”
Listen at 1:20:29
Kamala Harris is not viable with voters for another presidential run
“she is a non-starter with voters”
Listen at 1:21:58
Kamala Harris’s calculated communication style will fail in 2028
“It will not work in a 2028 run.”
Listen at 1:23:12
Zohran Mamdani’s affordability focus resonates with voters’ cost concerns
“the fact that he is relentless about saying, I know the city's too expensive for you to live in”
Listen at 1:25:56
The current Republican factional conflict is likely to shape 2028
“I think the current dynamic is a reflection of how things are shaping up for 2028.”
Listen at 1:27:57
Trump may leave office less popular than he is currently
“that Trump will not be as popular on the way out”
Listen at 1:28:18
Marco Rubio is more politically savvy than J.D. Vance
“Marco Rubio's just way more savvy about this stuff.”
Listen at 1:29:51
J.D. Vance’s position is worsening beyond his ability to recover politically
“every day that goes by, Vance is being put in a worse and worse position, one that he is not talented enough to work his way out of.”
Listen at 1:30:15
Sarah Longwell considers J.D. Vance personally repellent
“He is a repellent individual.”
Listen at 1:30:25
Republican voters view J.D. Vance negatively
“the Republican voters are pretty negative about him”
Listen at 1:31:28
Marco Rubio’s 2028 path depends on voters viewing Democrats as too far left
“I do think the path for Marco Rubio, though, would be really people saying, like, I think the Democratic Party is too far left.”
Listen at 1:33:50
The post-Trump Republican Party lacks a coherent policy platform
“When Trump is gone, it's not a party that has a policy platform.”
Listen at 1:35:50
Democrats face a historic opportunity against a weakening Republican Party
“historic generational opportunity to go on offense against a flatlining Republican Party”
Listen at 1:36:13
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
