Nonrival logo

Nonrival

Archives
Log in
Subscribe
September 8, 2026

AI insurance gaps, drone innovation, and Venezuela's oil gamble

Nonrival

Expert analysis from think tanks and academia, on public policy, economics, and technology.
Human experts. AI summaries.


Insurance Companies Are Quietly Becoming America's Most Powerful AI Regulators — by Refusing to Cover It

csis

  • State insurance commissioners have approved over 80% of insurer requests to exclude AI-related damages from corporate policies, meaning most businesses may soon find AI activities uninsurable — which, in practice, is nearly equivalent to those activities being banned.
  • The insurance industry can't price AI risk because it lacks basic data on loss frequency and severity, and can't verify what AI systems companies are actually running — so rather than setting standards for safe AI use, insurers are simply walking away from the market.
  • The author proposes four fixes: a federal AI incident database, a federal-state insurance working group, a government backstop for catastrophic AI losses, and independent auditors to verify AI risk claims — arguing that without these, the U.S. will have de facto banned AI deployment without ever debating it.

Trump's Venezuela Oil Deal Is Built on Shaky Legal Ground That Could Doom It

csis

  • The Trump administration struck a sweeping deal giving a two-year-old American company, NABEP, 100-year concessions over 65 billion barrels of Venezuelan oil — making it overnight one of the largest private oil companies in the world by reserves.
  • The deal has serious legal problems: Venezuelan law doesn't actually allow 100-year concessions, the interim government granting them lacks a democratic mandate, and the arrangement may violate Venezuela's constitution — all of which could invite future repudiation by an elected Venezuelan government.
  • The author argues the U.S. is repeating a historical mistake of taking shortcuts that prioritize production over legitimate institutions, and that lasting energy security in the hemisphere requires transparent, democratically grounded agreements, not opaque deals with a transitional regime.

Students Who Use AI to Do Their Homework Learn Less, New Research Finds

voxeu

  • Students who use generative AI tools to complete schoolwork show measurable learning deficits compared to those who don't, suggesting that the shortcuts AI enables come at a real cognitive cost.
  • The research, conducted in secondary schools, finds that AI use tends to substitute for rather than complement genuine learning — students get answers without building understanding.
  • The findings raise urgent questions for educators and policymakers about how to integrate AI into classrooms without undermining the learning process itself.

In the News

The US Job Market Is Holding Steady, but That Won't Settle the Fed's Debate Over Rate Hikes

employ_america

  • August added 162,000 jobs — well above expectations — and earlier months were revised upward, painting a picture of a labor market that's stable and no longer deteriorating.
  • Key stress indicators like short-term unemployment and the Black-white unemployment gap have improved, but real wage growth is negative and the gains may be slightly overstated due to seasonal quirks in government education employment.
  • With the labor market off the table as a concern, the Fed's September decision hinges on inflation data — and the deeper problem is that the usual mechanism by which rate hikes cool inflation (slowing demand, raising unemployment) doesn't obviously apply when inflation is being driven by tariffs, commodity shocks, and a rate-insensitive investment boom.

Analysis

Russia Has Turned a Cheap Iranian Drone Into a Case Study in Wartime Innovation — and the U.S. Can't Keep Up

csis

  • Russia acquired Iran's Shahed drone in 2022 and has spent four years relentlessly upgrading it — improving navigation, stealth, communications, and electronic warfare resilience — at a pace measured in weeks, not years.
  • The iteration model is striking: battlefield failures drive rapid redesigns, ad hoc hacks get taped onto drones mid-war and later formalized as production features, and two competing Russian factories run parallel development programs drawing on Chinese and Iranian components.
  • The article argues this represents a fundamental shift in how military advantage is won — not by building the most sophisticated system upfront, but by maintaining the organizational capacity to continuously improve under fire — a model the U.S. is not currently structured to match.

Also Worth a Look

  • AI in a fragmenting world: Why 'connector' economies can still gain (voxeu)
  • Big Tech's $1 Trillion AI Bet Requires a Productivity Miracle Bigger Than Any in History (wharton)
  • Uber's Personalized Pricing Isn't the Real Threat to Riders — Platform Market Power Is (microecon_insights)
  • A Police Officer Makes the Case for the License-Plate Cameras Americans Are Tearing Down (works_in_progress)
  • 25 Years After 9/11, Anti-Muslim Sentiment in America Is Worse Than It Was Right After the Attacks (pew)
  • A Weak Tropical Storm Just Dumped Nearly 30 Inches of Rain on East Texas — and a Bigger Threat Is Heading Toward Hawaii (yale_climate)
  • Three Pacific Hurricanes Are Churning Simultaneously for the First Time in 11 Years, and One Is Heading Toward Hawaii (yale_climate)
Don't miss what's next. Subscribe to Nonrival:
← Newer AI adoption gaps, pollution's toll on learning, and corrupt government Older → SEC retreat, ESG pushback, and heat risk for athletes
Powered by Buttondown, the easiest way to start and grow your newsletter.