Sep 15, 2026 · 22 min
AI warnings collide with a bond-market stress test
Should markets discount the AI apocalypse?
The episode asks whether frontier AI’s risks deserve a market discount—and whether rising Treasury yields could expose the boom’s financial vulnerabilities.
- 1AI leaders’ warnings may combine genuine concern, commercial self-interest, and uncertainty about catastrophic risks.
- 2Public ownership could improve disclosure, but executives may still shift responsibility for failures onto governments and taxpayers.
- 3A bond-market sell-off, including Treasury yields above 5 percent, could challenge the financing behind the AI investment boom.
Don't miss
The hosts connect AI’s financing needs to a bond market that failed to rally as a traditional safe haven, then ask whether yields above 5 percent could puncture the boom.
The brief
Katie Martin and Rob Armstrong examine warnings that autonomous, recursively self-improving AI could become dangerous, while questioning why industry leaders are sounding the alarm now.
The hosts weigh genuine concern against commercial self-interest, asking whether companies such as OpenAI and Anthropic should control frontier systems or disclose more as public firms.
Their accountability debate turns on a slippery “we”: AI executives call for guardrails, but may be asking governments and taxpayers to own the consequences.
Stocks wobbled after the warnings, but government bonds failed to rally as a safe haven; the hosts ask whether Treasury yields above 5 percent could threaten AI’s boom.
The episode closes with a lighter argument over loafers and tailoring, including a sharp critique of Scott Bessent’s suit fit and an FT column photograph.
What was said on this episode
12 statements · 5 positive · 6 negative · 1 mixed
AI’s frightening capability may benefit businesses developing it.
“this technology is so good that it's scary might be kind of good for the businesses involved”
Listen at 7:34
AI may offer strong growth with a small probability of destroying the world.
“95 percent or 98 percent strong growth and one to five percent destruction of the world”
Listen at 8:38
Investors can rationally go all-in on AI despite catastrophic tail risks.
“There is a case to be made. Go all in all the more so. Right. Because. The terrible outcomes don't matter from an investment point of view.”
Listen at 9:14
OpenAI and Anthropic should list publicly sooner to increase transparency.
“I think the sooner these companies list, the better. Get it into the sunshine as fast as possible.”
Listen at 12:40
Public listings would produce more disclosure about AI companies.
“We'd get more of them if these were publicly traded companies with obligations to disclose stuff that we want to know about.”
Listen at 12:55
AI leaders’ warnings diffuse accountability across companies, governments, and taxpayers.
“they are spreading accountability. They're smearing it across the industry, the government. taxpayers, everybody.”
Listen at 14:47
AI companies should accept responsibility if profitable technology harms people.
“We want those companies to say, if this technology, which we are going to profit from, gets out of control and hurts people, that's on us.”
Listen at 15:02
The U.S. 10-year Treasury yield rose above 5 percent.
“the 10-year yield... has popped up above 5%”
Listen at 16:32
The bond market is currently in serious trouble.
“the bond market is in a serious mess”
Listen at 17:14
Higher Treasury yields raise financing costs for the AI investment boom.
“The AI boom is an investment boom. Investment booms have to be financed. The 10-year yield reflects a higher cost of financing.”
Listen at 17:40
Higher oil prices and strong U.S. nominal growth are contributing to bond-market weakness.
“the bond market is responding to some combination of higher oil prices and strong nominal growth in the United States”
Listen at 18:09
People who move their arms should wear tailoring with high armholes.
“you should always wear tailoring with a high armhole if you're the kind of person who moves your arms”
Listen at 20:49
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.