DON'T MISS: A Strategic Framework for Defending American Economic Interests From The EU's Reckless CS3D Directive — Third and final in our article series on the EU directive CS3D and how it affects American business, in collaboration with Baron Public Affairs. Read and share!

Explaining Anthropic's Strategy Of Regulatory Capture

Whatever you may make of the changes on X dot com since its takeover by Elon Musk, it does remain the public square of the world, where some of the most important debates today are happening.

Here's a recent example: a phenomenon you have seen us discuss many times, whereby an AI company (or maybe a "research lab" possibly funded by an AI company) publishes some "research" showing how easily AIs can go rogue and do bad things. Except that if you look closely at the actual research, it simply reduces to: "If you tell an AI to play-act as a villain, it will play-act as a villain."

As one X user memorably put it: "I wrote 'I am alive' on a piece of paper, and placed into a photocopier. What I saw next has shocking implications."

This seems strange: why would companies go out of their way to advertise their products as dangerous?

And the answer is simple: regulatory capture.

These companies are trying to create an environment where they can steer AI regulations that prevent upstarts from one-upping them, in a highly competitive marketplace, where the leaders can raise massive amounts of money (which they can then, unusually for small, new companies, spend on lots of public affairs teams) but it's still conceivable that some team in some garage somewhere might come up with an even better mouse trap.

During the Biden administration, pretty much every major AI company played this game. In the Trump administration, most companies have understood that the US has decided to actually have a free market when it comes to innovation. (OpenAI, in particular, has pivoted admirably.)

Anthropic, however, was the worst offender under the Biden administration and has continued with the same strategy under the Trump administration.

All of this has come to light. Jack Clark, co-founder and Head of Policy at Anthropic (how often do you see a Silicon Valley startup whose co-founder is "Head of Policy"?) published an essay titled "Technological Optimism and Appropriate Fear." The essay is worth reading and perhaps taking seriously, but the title gives the flavor of the thing. AI is dangerous!

This prompted a sharp riposte from AI Czar David Sacks: "Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering. It is principally responsible for the state regulatory frenzy that is damaging the startup ecosystem."

You will have gathered from the prolegomena of this article that we completely agree with Sacks. Anthropic supported California's terrible AI bill, for example.

Of course, Sacks' tweet created a firestorm, with angry denials from Anthropic in particular.

The best response, however, was from Neil Chilson, a former FTC chief technologist and head of AI policy at the Abundance Institute. In a long thread, he basically detailed how Sacks (and your correspondent) are totally correct, pointing out that if you parse Anthropic's own statements carefully, they admit this frankly.

Chilson, who attended Clark's keynote at The Curve conference in Berkeley, which was the basis for his essay, revealed that Clark's essay omitted key details from a Q&A session. In the Q&A, Clark advocated for a "major government regime with teeth," including pre-deployment testing, expanded liability, and mandatory API access. More provocatively, he described transparency laws as a "first step" to "scare people enough to draw attention and create the political will" for stricter rules.

Chilson also quoted Anthropic CEO Dario Amodei from a New York Times op-ed: transparency helps Congress "understand how the technology is developing" to decide on further steps. This, Chilson argues, leads inexorably to pre-approval regimes, which historically foster capture. In other words, Anthropic is actually clear that their demands for "transparency" are just a first step towards a more restrictive regime, and specifically a pre-approval regime.

He reprised the well-known "Bootleggers and Baptists" theory, casting Anthropic, with great charity, in the sincere "Baptists" camp, but pointing out that even if the baptists are sincere, cynical "Bootleggers" still get to create a regime that entrenches incumbents.

"If that isn’t a regulatory capture strategy based on fear-mongering, then what is it?," Chilson asks rhetorically. Indeed.

Policy News You Need To Know

DON'T MISS: A Strategic Framework for Defending American Economic Interests From The EU's Reckless CS3D Directive — Third and final in our article series on the EU directive CS3D and how it affects American business, in collaboration with Baron Public Affairs. Read and share!

#Germany #Manufacturing #GreenNewScam — Germany's machinery sector, a cornerstone of the nation's economy contributing 7% to GDP and employing over one million people, is facing a severe downturn, with output plummeting 22% since pre-COVID levels and revenues projected to drop another 5.6% in 2025, according to a PwC report and VDMA forecasts warning of up to 10% further production declines amid record-low business confidence (via Apollo News). High energy costs are cited as primary culprits, leading to 12,000 job losses, a 22% surge in insolvencies, and the lowest capacity utilization in five years. Don't listen to Greta!

#Trade #Chyna — The Journal has a long reported piece by Lingling Wei on what it's calling "China's New Strategy for Trump." The strategy apparently combines aggressive retaliation with strategic concessions. The approach involves hitting back harder than the US when faced with economic pressure while offering deals on high-visibility issues Trump personally prioritizes, such as allowing the sale of TikTok. The most significant example came on October 9 when China unveiled sweeping restrictions on rare-earth exports as leverage before the leaders' scheduled meeting next week in South Korea. Xi, who privately dismissed TikTok as "spiritual opium," according to the report, believes Trump is fundamentally transactional rather than ideological and can be influenced through this "soft and hard" approach, while traditional China hawks within the Trump administration have reportedly been marginalized. The strategy has shown mixed results: while it initially emboldened Xi after market chaos from rare-earth restrictions in April, the October announcement triggered threats of 100% tariffs from Trump and forced Beijing to promise "prudential and moderate" implementation of the controls. It's all a parlor game and Kremlinology. We are still trusting the plan.

#IndustrialPolicy — The Export-Import Bank announced its participation in President Trump's American AI Exports Program. Operating through its China and Transformational Exports Program, a Congressional mandate signed by President Trump, EXIM is deploying comprehensive financing tools including export credit insurance, working capital guarantees, and direct loans to accelerate US AI infrastructure, hardware, and software exports globally. The initiative, launched under Executive Order 14320 and coordinated through the Economic Diplomacy Action Group alongside OSTP and the Departments of Commerce and State, aims to be a whole-of-government approach to maintaining American technological leadership against strategic competitors. This is just good old-fashioned industrial policy. Welcome to the MAGA future!

#PublicHealth — This headline from Blaze Media says it all: "Gay-spread monkeypox is back. Watchdog asks policymakers to drop the political correctness." Reasons of political correctness cause public health officials to hide how certain diseases are spread through the gay community, which makes response less effective. That's bad.

#Crypto — Bloomberg: "JPMorgan Chase & Co. plans to allow institutional clients to use their holdings of Bitcoin and Ether as collateral for loans by the end of the year in a significant deepening of Wall Street’s crypto integration."

#ThinkTanks — RAND have announced a layoff of 11% of their staff. The whimsical folks at DARC comment: "Structural shifts are underway that mean that 2025 will be the high water mark for the grand edifice of postwar defense and foreign affairs think tanks once considered nearly unshakable in their influence over US policy. The field is increasingly wide open."

#LawAndOrder — Zohran Mamdani has publicly committed to closing Rikers prison, even though it will mean releasing 7,000 criminals. He is basically Bane from the Batman movie. Of course, this has implications not just for New York: he represents the future of the Democratic Party…

#Science #QuantumComputing — Google have said they have achieved a significant milestone in quantum computing with their Willow chip, demonstrating the first verifiable quantum advantage in a practical algorithm. This breakthrough, published in Nature, represents more than incremental progress: the algorithm ran 13,000 times faster than classical supercomputers while maintaining verifiable, repeatable results. This ability represents a crucial threshold for transitioning quantum computing from theoretical curiosity to practical tool. The immediate applications center on molecular modeling, particularly enhancing Nuclear Magnetic Resonance spectroscopy for drug discovery and materials science, where quantum computers can reveal atomic interactions previously inaccessible to observation. However, the broader implications warrant careful consideration: as quantum computing matures toward error-corrected, large-scale systems, it poses profound challenges to contemporary cryptographic infrastructure, including the encryption protocols securing financial systems, communications networks, and blockchain technologies. Current public-key cryptography, upon which cryptocurrencies fundamentally depend, becomes vulnerable to sufficiently powerful quantum computers capable of efficiently solving problems like integer factorization, though we have to note that today's demonstration, while significant, remains distant from cryptographically-relevant quantum computing. Still, every expert we know seems to think this is a real and significant breakthrough, and the biggest in a very long time.

Friday Essays

DON'T MISS: A Strategic Framework for Defending American Economic Interests From The EU's Reckless CS3D Directive — Third and final in our article series on the EU directive CS3D and how it affects American business, in collaboration with Baron Public Affairs. Read and share!

This week, there are two absolutely must-read essays.

The first is the latest essay by Peter Thiel, and his aide Sam Wolfe, in the latest issue of First Things. (We've been informed it has come out for a few weeks but for some reason only caught our attention now.) The essay is on his most recent obsession: the Antichrist. Like every Peter Thiel intellectual venture, the essay defies easy description. It opens with a tour-de-force Straussian reading of Francis Bacon's New Atlantis, moves on to Jonathan Swift and, believe it or not, comicbooks and manga. Just read it.

The second essay is perhaps even scarier than the Apocalypse. It is titled "The Goon Squad," by Daniel Kolitz, in Harper's, and it is about the culture of "gooning." (By the way, what is it with Harper's that it seems to hum outside public view as an old left magazine, and then every six months or so, come up with an absolute earthquake of a must-read essay?) If you don't know what "gooning" is, God bless you. We will let Kolitz describe it: "Gooning is a new kind of masturbation. More precisely, a new kind of masturbation at the heart of an internet-based, pornography-obsessed, Gen Z–dominated subculture every bit as defined and vibrant as the hippies or punks in their prime. The act itself resembles 'edging'—repeatedly bringing oneself to the point of climax without actually climaxing. But gooning is more goal-oriented and more communal. The gooner goons to reach the 'goonstate': a supposed zone of total ego death or bliss that some liken to advanced meditation, the attainment of which compels them to masturbate for hours, or even days, at a time." Yes, this is the next stage of porn's ongoing devouring of the brains of the youth, and it is just as bad as you think. We cannot look away.

Incidentally, we would be remiss if we did not take this opportunity to flog our very old, but in retrospect perfectly vindicated, piece from 2019, making the case for the existence of porn addiction, the fact that it rewires the brain, and that it is essentially destroying society: A Science-Based Case for Ending the Porn Epidemic.

Many other good essays to mention, of course.

For example, the inimitable philosopher of neoreaction, Curtis Yarvin, has given a very long interview at the Carson J Becker Substack on the history of the 20th century and of World War II. It is, of course, a revisionist history, but, as always with Yarvin, a very interesting and very learned one, whether you agree or disagree.

Over at Compact, the UCLA sociologist Gabriel Rossman, always a very perceptive writer, laments the effects that AI has on rotting students' brains. "That improving AI is a substitute for labor is widely appreciated, and indeed, megalomaniacal mad-scientist visions of mass white-collar unemployment is part of the pitch decks shown to investors. What is less widely appreciated is that even if the technology stopped improving tomorrow, it would still be an increasingly good substitute for human capital. This is because it is already capable of giving human beings, and especially young people, the choice to idle in stupidity and ignorance."

At The Atlantic, Julia Steinberg, who writes for magazines like Arena and Palladium, writes about the conservative renaissance at Stanford. "What outsiders might not understand is that, at least in my experience, the appeal of conservatism on campus today isn’t really about Donald Trump or Trumpism, or any other set of ideological beliefs. … The Review staff included MAGA diehards, traditional Catholics, anti-Trump neoconservatives, isolationists, anti-identity-politics liberals, Luddites, and (in my case) techno-capitalists, all challenging one another’s ideas. Some of us voted for Trump; some of us did not. Still, most of us were excited when he won."

DON'T MISS: A Strategic Framework for Defending American Economic Interests From The EU's Reckless CS3D Directive — Third and final in our article series on the EU directive CS3D and how it affects American business, in collaboration with Baron Public Affairs. Read and share!