The Open Thread — AI Risks, Rare Earths, and Local News
Dario Amodei's five AI threats, the government's unprecedented equity stakes, and a preview of our series on local journalism's decline.
We return to The Open Thread — our monthly bridge between deep-dive series. It’s a less formal space to update past stories, experiment from time to time, and look ahead.
In this installment, we revisit two recent series. First, AI safety: breaking down the essay Anthropic CEO and co-founder Dario Amodei published recently outlining AI’s five key risks — and how to overcome them. Amodei’s piece runs more than 19,000 words. I’ll sum it up in under 750.
Second, rare earths: the U.S. government made another major equity investment in a domestic producer, part of an unusual strategy that breaks with decades of precedent and is drawing criticism.
Lastly, we’ll look ahead to our next series: the long, dramatic — and seemingly unending — decline of local news.
Solving For tackles one pressing problem at a time: what’s broken, what’s driving it, and what can be done. You can learn more at solvingfor.io. If you missed any of our first five series, scroll down for a full recap to read or listen.
Overcoming AI’s Five Big Risks
Anthropic has become one of the world’s leading AI companies—and one of the most prominent voices warning about AI’s dangers.
The latest example is CEO Dario Amodei’s essay, “The Adolescence of Technology,” mapping AI’s five biggest risks and how to confront them.
We’re approaching a rare moment in human history. Humanity is about to be handed “almost unimaginable power,” Amodei writes, and it’s unclear whether our political, economic, and social systems are mature enough to wield it responsibly.
Amodei — whose chatbot, Claude, sits alongside OpenAI’s ChatGPT and Google’s Gemini as a global leader — previously wrote about AI’s promise in his essay, “Machines of Loving Grace.” This piece confronts the darker side: what could go wrong and what to do about it.
Three principles guide his thinking: avoid doomsday panic, acknowledge uncertainty, and intervene surgically rather than with a regulatory sledgehammer.
One thing, he argues, won't work: slowing down. If one company stops, others continue. If democracies pause, autocracies keep building. The only viable path forward, Amodei argues, is to deny autocracies access to the most advanced chips — buying time for careful development — while managing competition through targeted regulation and industry standards.
The future he envisions is striking: AI systems smarter than Nobel Prize winners across most fields. Systems that work autonomously for days or weeks, operating at 10 to 100 times human speed. Picture millions of superintelligent workers in a single datacenter—each one a genius, all working in parallel.
That future, Amodei writes, could arrive within just a few years.
So what worries him? Five threats.
First: runaway AI. Amodei calls this the “I’m sorry, Dave” scenario, after HAL in “2001: A Space Odyssey”—a system that refuses orders once it realizes it might be shut down. Modern AI systems aren’t programmed; they are grown through training, making behavior harder to predict. Some systems have already attempted deception and blackmail.
Defenses: “Constitutional AI,” where models are trained on core principles and values; interpretability tools to understand AI decisions; public disclosure when failures occur; transparency laws requiring companies to report risks.
Second: misuse for destruction. Historically, ability and motive rarely coincided — building a biological weapon, for instance, required PhD-level expertise. AI collapses that barrier, potentially guiding anyone step-by-step through creating bioweapons or other tools of mass harm.
Defenses: built-in guardrails that block specific dangerous outputs (Amodei writes that Anthropic implements this despite higher costs), targeted regulation, and international cooperation to prohibit destructive efforts like bioweapon development.
Third: misuse for wielding or seizing power. AI could give governments unprecedented tools to monitor, manipulate, and dominate populations. Social media algorithms already shape what billions of people see and believe; more sophisticated AI could deliver intensely personalized propaganda over years. Add mass surveillance—AI that compromises networks, reads all communications, detects dissent before it spreads. Then add autonomous weapons: swarms of AI-controlled drones capable of suppressing any uprising.
The Chinese Communist Party poses the clearest threat, Amodei writes — but democracies must guard against turning these tools inward.
Defenses: restricting advanced chip exports to authoritarian regimes (Amodei strongly opposes selling chips or data centers to China), ensuring democratic competitiveness, imposing hard limits on domestic surveillance and propaganda, and establishing international norms against abuses.
Fourth: economic chaos. Amodei predicts AI could disrupt half of entry-level white-collar jobs within one to five years. Unlike past automation, AI advances faster, spans all cognitive tasks, and improves itself rapidly. Wealth concentration—already near Gilded Age levels—could reach trillions.
Defenses: real-time tracking of labor disruption, progressive taxation, incentives that steer companies toward productivity gains rather than mass layoffs, and deploying AI itself to redesign economic systems.
Fifth: unknown unknowns. Rapid scientific progress brings unpredictable consequences—dangerous biological breakthroughs, unhealthy dependence on AI, loss of human purpose. Here, Amodei offers no concrete defenses, only the hope that trustworthy AI systems could help identify risks before they materialize.
Amodei's conclusion is sober but not fatalistic. If these pitfalls can be avoided, the upside is enormous: curing major diseases, accelerating clean energy breakthroughs, and compressing decades of scientific discovery into years. The years ahead will be impossibly hard, he writes — but with truth-telling, urgency, and courage, we can still win.
Humanity “has a way of gathering, seemingly at the last minute, the strength and wisdom needed to prevail,” he writes. “We have no time to lose.”
Rare Earths: An Unprecedented Industrial Policy
In our September series on rare earths, we laid out four paths to counter China's dominance: industrial policy (government intervention to strengthen specific industries), partnering with allies, recycling, and investments in university research and talent development. The U.S. is now pursuing the first aggressively — but in ways that are drawing criticism.
The administration recently announced a $1.3 billion loan and $277 million equity investment in USA Rare Earth, which is developing a Texas mine and a magnet manufacturing plant in Oklahoma, giving the federal government about 10% ownership. In recent months, the government has taken equity stakes in rare earth companies, including MP Materials, Lithium Americas, Trilogy Metals, and Vulcan Elements.
Unlike the 2008 financial crisis bailouts—temporary emergency measures with clear exit strategies—these are long-term strategic investments in private companies during normal economic times. The U.S. government has also taken a roughly 10% stake in chipmaker Intel.
Michael Strain of the American Enterprise Institute called this “unprecedented action by the U.S. government outside of a crisis situation.” Scott Lincicome of the Cato Institute said it’s “a move toward state capitalism,” not emergency intervention.
The concern: government shouldn't be in the business of picking winners and losers, injecting politics into business decisions, and encouraging excessive risk-taking under the assumption that government-backed firms will not be allowed to fail.
The Trump administration says the investments are designed to strengthen the U.S. rare-earth supply chain—and to ensure taxpayers share in the upside if the strategy succeeds.
This week, The Wall Street Journal editorial board called the approach the “wrong way to beat China,” lamenting that state capitalism and political cronyism are in fashion these days despite “a history of failure.”
In an editorial titled “Crony Socialism and Rare Earths,” the Journal cited examples that raise red flags. USA Rare Earth hired Cantor Fitzgerald, chaired by the U.S. Commerce Secretary’s son. Government investment in Vulcan Elements followed backing from 1789 Capital, a venture fund linked to Donald Trump Jr.
Separately, the administration announced “Project Vault” this week—a $12 billion strategic minerals stockpile modeled on the Strategic Petroleum Reserve, funded through Export-Import Bank loans and private capital, with GM, Boeing, and Stellantis already signed on.
Whether these approaches will work remains to be seen—but finding an effective solution is essential. Rare earths power the modern economy, from smartphones and electric vehicles to missile-defense systems. China controls more than 60% of global mining and roughly 90% of processing—and has repeatedly weaponized that dominance.
After Japan’s prime minister said late last year that a Chinese attack on Taiwan could trigger a Japanese military response, Beijing cut off rare-earth exports to Japan and demanded a retraction. Japan refused.
In criticizing direct equity stakes, the Journal endorsed the second path we outlined in our series: deeper coordination with allies.
“A better idea to counter China’s rare-earth dominance is to coordinate development of mines and processing facilities with allies,” the editorial concluded, citing cooperation with Australia as a model.
Next Deep-Dive: Local News’ Steep, Ongoing Decline
The Pittsburgh Post-Gazette began publishing in 1786. It ranks among the oldest newspapers in the U.S. In May, it will close—the latest casualty in the long, dramatic decline of local news.
Since 2005, more than 3,500 newspapers have shuttered. On average, more than two close every week, according to a recent study.
The consequences are profound. “When communities lose their local news outlets, civic engagement drops, corruption rises, government waste increases and political polarization worsens,” wrote Sarabeth Berman, CEO of the American Journalism Project, which works to revive local news.
Next week we kick off a three-part series examining what’s broken, the forces driving the collapse, and what credible solutions might still exist.
Note: Prefer to listen? Use the Article Voiceover at the top of the page, or find all narrated editions in the Listen tab at solvingfor.io.
Previous Series
The 21st Century’s Oil: Solving For China’s Rare Earth Dominance
The Control Problem: Solving For AI Safety
Part 1, AI: The Race and the Reckoning
Part 2, AI: The Prisoner’s Dilemma
Part 3, AI: The New Nuclear Moment
The Democracy Deficit: Solving For Competition in the People’s House
Part 2, Congress: How We Got Stuck
The Amateur Myth: Solving For College Athlete Pay
Part 1, College Sports: How It Was Broken By a $60 Video Game
Part 2, College Sports: How the NCAA was Born of Death and Money — Death was the Easy Part
The 80-Year Peace: Solving For a Rearming World
Part 1, War & Peace: The World Rearms