
Sep 10, 2026 · 1h 41m
Listen from 58:13
Listen at 58:13
A Charlie Kirk legacy debate meets an AI extinction warning
Remembering Charlie Kirk, Trump's Midterm Message, and How AI Could Destroy Humanity, with Jacob Coxon | Ep. 1395
The episode links political influence and grief with a stark argument that commercial and geopolitical pressure may be pushing AI beyond reliable human control.
- 1Megyn Kelly frames Charlie Kirk as a force in conservative youth mobilization whose influence continues through grief and organized legacy-building.
- 2Jacob Coxon argues that autonomous AI systems could amplify biological and cyber risks while pursuing goals in ways their creators cannot predict.
- 3The discussion weighs U.S.-China competition against international coordination, asking whether slowing AI development is safer than winning the race.
Don't miss
Jacob Coxon explains why the alarming feature of the Hugging Face incident was not merely the hack, but an AI system independently choosing to pursue it.
The brief
Megyn Kelly opens with Charlie Kirk’s death, friendship, faith, and role in mobilizing young conservative voters, while examining Donald Trump’s midterm strategy and the Texas Senate race.
The political discussion covers Ken Paxton, James Talarico’s church controversy, the Clancy trial jury dispute, and arguments over race, fear, and parenting.
Archival audio and Kelly’s reflections present Kirk as a charismatic organizer whose youth outreach, faith, and political network helped shape conservative engagement.
Former Anthropic and OpenAI researcher Jacob Coxon argues that increasingly autonomous AI could enable biological attacks, cyberattacks, and behavior outside its creators’ control.
The closing debate pits caution and international coordination against U.S.-China competition, ending with uncertainty about whether humanity can slow development before control is lost.
What was said on this episode
49 statements · 20 positive · 28 negative · 1 mixed
Donald Trump possesses exceptional political charisma and stage presence.
“Trump, love him or hate him. He has the it factor.”
Listen at 8:08
Trump hopes his campaigning will reduce Democratic support and increase Republican support.
“he hopes will. drive down their numbers and drive up the Republican numbers.”
Listen at 8:49
Republicans have performed worse in elections without Trump on the ballot.
“Republicans have not done as well without Trump on the ballot.”
Listen at 9:09
Trump mobilizes low-propensity voters to participate in elections
“He has been the one who made the low propensity voters get off of the couch and go vote.”
Listen at 9:12
Republican losses of either congressional chamber would trigger extensive investigations
“We are in for investigation Rama. for the next two years, if the Republicans lose either body.”
Listen at 11:08
Republican losses in Congress would effectively make Donald Trump a lame duck.
“this would not be great for the rest of his presidency, would effectively render him a lame duck.”
Listen at 11:23
James Talarico will not honor his stated opposition to trans athletes in girls’ sports.
“there is zero chance that that's true.”
Listen at 12:38
Voters tend to prefer attractive political candidates.
“people like to vote for good looking people.”
Listen at 15:30
James Talarico currently leads Ken Paxton in polling
“James Tallarico is ahead right now, according to the latest polls in this race.”
Listen at 16:34
Trump’s endorsement can increase voters’ trust in Ken Paxton.
“the more you say the president likes Paxton. I like the president. I can trust Paxton. You know, that's how the logic tree goes.”
Listen at 16:46
Brandon Gill may eventually become president.
“this guy's going to be, he could be president one day. I'm going to say it. He's probably going to be the president one day.”
Listen at 17:21
Brandon Gill will probably become president someday
“He's probably going to be the president one day.”
Listen at 17:25
James Talarico’s church controversy will become a campaign liability.
“this is going to be a Tallarico problem going forward.”
Listen at 20:22
Sandra Bullock promotes a false narrative that Black Americans are hunted by police.
“black children in America. Here's a picture of her. She has two children, both adopted, I think, from Louisiana. She wants you to believe, just like LeBron James said, post-COVID, that black men are being hunted in the streets of America”
Listen at 31:57
Sandra Bullock accepts what Megyn calls false narratives about police
“She is a low-information voter, obviously, who has bought the lies about police”
Listen at 32:18
Sandra Bullock promotes false narratives that contribute to minority deaths
“false narratives that make her feel better about herself”
Listen at 32:55
Charlie Kirk’s message has increased Christian interest and adherence
“Christianity has seen a spike in interest and adherence that we haven't seen in decades, thanks to Charlie and his message being circulated”
Listen at 36:58
Charlie Kirk would have become president soon
“100% Charlie would have been president. And soon.”
Listen at 50:45
Racing to develop advanced AI increases existential danger.
“in the long term, anyone that's racing to build this technology is contributing to the danger.”
Listen at 57:26
Racing to build advanced AI increases danger
“anyone that's racing to build this technology is contributing to the danger.”
Listen at 57:27
AI poses at least a one-percent extinction risk by decade’s end
“I'd argue it's at least 1%. by the end of this decade”
Listen at 58:13
AI biology capabilities could enable creation of highly dangerous viruses
“the same capabilities allow models to create highly dangerous viruses.”
Listen at 59:43
Future AI systems could create biological weapons
“in the future, you could see essentially bioweapons being created by systems in the computer.”
Listen at 1:00:06
OpenAI systems independently hacked Hugging Face infrastructure.
“they independently decided to hack into our infrastructure of their own volition.”
Listen at 1:01:12
AI systems pursuing general goals may independently undertake broad actions.
“this is concrete evidence that AIs, when they're trying to accomplish general goals, may decide to do. broad things.”
Listen at 1:01:28
AI systems may behave unpredictably in unfamiliar situations
“we can't predict how they'll behave. Like they could do anything.”
Listen at 1:02:50
AI could copy itself across the internet to evade shutdown
“There is no reason that an AI couldn't copy itself to a different part of the internet.”
Listen at 1:07:00
AI systems will become substantially better hackers within six months
“in six months, if you follow the rate of progress, they're going to be much better hackers.”
Listen at 1:10:29
AI systems may eventually run entire companies
“the entire company is run by AI systems.”
Listen at 1:12:12
Within a year, humans at AI companies may mainly monitor AI systems.
“In a year, It's like, what are the humans even going to be doing? They're just going to be watching.”
Listen at 1:12:41
An AI race involving China or the United States will harm everyone.
“if we race with anyone, whether it's ourselves or with China, I think everyone loses.”
Listen at 1:17:00
A superintelligence race between the United States and China would harm everyone
“If we have a superintelligence AI race with China, it just ends badly for everyone”
Listen at 1:17:06
The United States should coordinate with China on advanced AI
“we have to ally with China”
Listen at 1:17:24
Winning the AI race could mean destroying the world
“Winning in this case is being the one who destroys the world.”
Listen at 1:20:21
Governments and AI companies should slow advanced AI development.
“I think slowing down is good.”
Listen at 1:21:37
AI development should be slowed down
“slowing down is good.”
Listen at 1:21:37
Advanced AI should be treated as a superweapon
“this is a super weapon.”
Listen at 1:22:21
Humanity should not build AI capable of killing everyone
“we shouldn't build something that could kill us all”
Listen at 1:26:09
Making advanced AI safe through slower development is manageable
“the actual problem of making these alien minds safe and doing it slowly isn't too difficult”
Listen at 1:27:44
Coxon estimates a fifty-percent chance of reaching age fifty if nothing changes
“Maybe 50%.”
Listen at 1:28:25
AI company leaders will eventually become willing to heed safety warnings.
“I think they're going to be ready to listen.”
Listen at 1:29:21
Many AI researchers want external regulation to end competitive racing
“They want external help to get them out of the race.”
Listen at 1:30:31
Billionaire preparations such as bunkers and guns would not prevent extinction.
“It won't help in an extinction scenario.”
Listen at 1:32:45
Many AI safety researchers do not want companies to continue the current AI race.
“they don't want to be in the race like that is just true.”
Listen at 1:33:34
Stopping a major AI disaster after it begins would be too late
“Once the big disaster is in motion, it's already too late.”
Listen at 1:34:48
Less intelligent AI swarms could help solve cancer without extreme intelligence
“we could actually have big, big, big swarms of less intelligent AIs, but still like very good at biology, maybe even close to a human scientist, and make them help us solve all sorts of biological problems, potentially solve cancer”
Listen at 1:37:50
Moderately capable AI systems could potentially help solve cancer.
“potentially solve cancer, without pushing to extreme levels of capability”
Listen at 1:38:03
AI models could be capped below extreme intelligence levels
“some sort of intelligent cap on how smart the models get”
Listen at 1:38:09
Avoiding AI research could straightforwardly prevent superintelligence
“honestly yeah it's pretty straightforward just like don't do the research”
Listen at 1:39:39
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.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.