⚡ Wargaming an AGI cyber surprise
Also: A Quick Q&A with policy analyst Jordan McGillis on whether California can build its way out of stagnation—literally
My fellow pro-growth/progress/abundance Up Wingers in the USA and around the world:
If we're only a year or two away from broad, human-level AI—artificial general intelligence—then pretty much all of our bandwidth should be devoted to analyzing its socioeconomic, geopolitical, and national security consequences. Yet even if you think AGI anytime soon is highly unlikely, its potential impact still merits considerable thought in the here and now.
I’m glad, then, that the RAND Corporation is running “Day After AGI” exercises to explore policy problems unique to AGI or the advanced capabilities that might emerge on the path to AGI. Its latest report documents the results of a series of tabletop exercises in 2025 that simulate what an AI-driven cyber confrontation between America and China might look like. Six sessions brought together RAND analysts, outside experts, and, in some cases, former senior national-security officials. Participants were asked to role-play members of the U.S. National Security Council’s Principals Committee, the cabinet-level body that advises the president during emergencies.
RAND’s fictional scenario, dubbed “Cyber Surprise,” begins with an unsettling discovery. U.S. intelligence learns that China has developed an advanced, cyber-AI system, known as LING MAO, capable of automating software engineering tasks such as code analysis, generation, and vulnerability repair. As the system comes online, long-standing American access to Chinese digital networks suddenly begins to disappear. Systems that once contained exploitable weaknesses are patched at unprecedented speed.
Over the coming week, the crisis escalates. A wave of cyberattacks strikes key sectors of the U.S. economy, including manufacturing, banking, and telecommunications, while Taiwan experiences crippling cyber disruptions ahead of a major typhoon that threatens semiconductor production and communications systems. Evidence points toward Beijing, but not conclusively. The attacks could be the work of the Chinese state, an AI-enabled capability operating beyond its operators’ control, or some third party exploiting the confusion.
What did RAND learn from how the scenario played out? Participants showed a striking willingness to escalate. Faced with the possibility that China had gained a decisive cyber advantage, many argued for aggressive countermeasures—sabotaging the rival system or exploiting whatever cyber access remained before it disappeared. Others even raised the possibility of intelligence operations aimed at acquiring the model itself. (Call it the Firefox Gambit.) Several participants framed the situation in “use-it-or-lose-it” terms. The fear was not merely of a temporary setback but of a lasting shift in the balance of power, one that could justify taking risks despite the possibility of retaliation.
At the same time, players felt largely blind. They lacked reliable intelligence about the rival system’s capabilities and about how far comparable American models had progressed. Even the basic question of attribution remained unresolved: The attacks might have been launched by China, by an AI operating outside its operators’ control, or by some other actor exploiting the confusion. Stymied by the fog of digital conflict, strategic decisions quickly turned into educated guesses.
The exercises also exposed a planning gap. Participants repeatedly argued that governments would need clearer playbooks before a crisis of this kind arrived—guidance for protecting critical infrastructure, managing economic disruption, and coordinating responses with allies. Without such preparation, policymakers would be forced to improvise under extreme pressure.
They would also have to improvise at great velocity. In RAND’s scenario, a breakthrough in cyber-AI quickly triggers geopolitical tension. Systems that automate vulnerability discovery and software engineering would accelerate both offensive and defensive cyber operations. What once unfolded over years could soon happen in months. It’s not hard to imagine how pre-AI institutions would be brittle under such conditions.
A destabilizing AI system needn’t be science fictional to automate conflict faster than rivals can respond. That possibility, rather than distant visions of superintelligence, is the one defense planners may need to worry about first—and sooner rather than later.
Q&A
🐻🏗️ California Forever … Or Ever?: A Quick Q&A with policy analyst Jordan McGillis
Goldman Sachs trader turned CEO Jan Sramek is guiding an ambitious project: build an entirely new city from scratch in Solano County, California. California Forever is built on the idea that the state of California, more specifically the Bay Area, was once a hub of industry and innovation that has since stagnated. The vision: a new city where affordable housing, plentiful jobs, thriving industry, and a walkable cityscape coexist. Sound like a utopia? Not all of the project’s neighbors have bought in.
In his recent City Journal article, “A New City of Dreams?,” Jordan McGillis captures the many mixed feelings circulating on the ground throughout Solano County. I asked McGillis a few quick questions about the attitudes surrounding the project, its potential pitfalls, and how hopeful we might be about California Forever’s success.
McGillis is a fellow at the Economic Innovation Group. He was previously the economics editor of City Journal, policy analyst at the Manhattan Institute, and deputy director of policy at the Institute for Energy Research. He is a Novak Journalism Fellow with the Fund for American Studies and is the author of his own Substack.
It has a bit of a utopian mystique to it, but the planners at California Forever seem genuinely motivated by California’s horrible failure to provide comfortable residential areas to median-earning families.
1/ Is the NIMBY-ism around California Forever generational? Does it seem that younger people are more amenable to it?
I think it can be interpreted as generational, with younger people being more enthusiastic about the project and older people being more tepid. I talked, for instance, with a Gen X mother in the outer eastern stretch of Solano County, in a town called Rio Vista, about it late last year. She’s lived there for two decades and hadn’t the slightest appetite for the change to the town that California Forever is already bringing, through its land acquisitions alone. Her Gen Z son, though, had this response when I asked him: “More jobs, more women. I like it.” That was striking, so I used the anecdote to round out my City Journal story.
That said, I think age might be a confounding variable, with ownership status being the more pertinent factor. Homeownership tends to correspond with age, but I know plenty of older Californians who find themselves outside of the state’s ownership class and are just as fed up with the status quo as younger Californians and just as ready for some dynamism.
2/ Will the kinds of people who might want to move to a new experimental, Silicon Valley-driven town like this one be attracted to the kinds of jobs Sramek wants to promote in the area?
It has a bit of a utopian mystique to it, but the planners at California Forever seem genuinely motivated by California’s horrible failure to provide comfortable residential areas to median-earning families. If they can create that sort of environment, a lot of people will just see it as a more affordable, livable option than the core Bay Area. I think it could become analogous to the Inland Empire’s relationship with Los Angeles. If you can get a solid job and a house you can afford, you’re not going to be too fixated upon the project’s origins.
The best argument—one that I think is proven in China’s megacity of Shenzhen—is that manufacturing works best when there are tight design-prototype-iteration loops.
3/ Is there a mismatch between the job base envisioned and the proposed population size?
Yes. I’ve highlighted this as the great paradox of the “Reindustrialize” movement and the techno-populist alliance. A significant portion of America’s techno-optimists (or accelerationists, if you want to use that term) has allied itself with right-wing populism. They have a shared and justified enmity against the anti-industrial elements of the political left, but don’t have a shared vision for what triumph over it looks like. The populists see manufacturing as the wellspring of middle-class prosperity by way of mass employment; the techno-optimists are racing toward the goal of the human-free “dark factory.”
Maybe a synthesis can be reached with California Forever and a proliferation of small manufacturers that require medium-skill technicians. I’m not sure though. More likely, I suspect, is that the modal resident of California Forever will work elsewhere in Solano County or have a long commute to either the East Bay or the Sacramento area.
4/ What’s the clearest economic case that firms will relocate to this new city rather than existing metros competing for similar industries?
The best argument—one that I think is proven in China’s megacity of Shenzhen—is that manufacturing works best when there are tight design-prototype-iteration loops. The Bay Area is the world’s strongest technology design hub. Siting manufacturing within a short drive makes sense if the endemic California labor and regulatory cost issues can be overcome. (The labor issues might be overcome by the dark factory, which, again, is the paradox of the Reindustrialize movement.)
In a real sense the project rests on the premise that the Bay Area won’t liberalize its land-use policies. In the very long term, though, there’s no reason America shouldn’t have larger conurbations.
5/ Aside from whether members of Solano County would want to be absorbed into the Bay Area, how might Bay Area residents theoretically feel about absorbing this neighbor?
Right now there’s essentially a giant rural gap in the center of Solano County. At the county’s western, Bay-facing edge, Vallejo is definitely part of the Bay Area. On the other side of that gap is Rio Vista, which has a strong Delta-inflected culture, making it more like Stockton than San Francisco. I don’t think Bay Area residents will have much of an opinion on California Forever. Angelenos don’t think about San Bernardino much, to draw that parallel again.
6/ Is California Forever’s success ironically dependent upon the rest of the state’s bureaucratic and political stagnation?
The project arose a decade ago on account of Sramek’s correct observation that the Bay Area doesn’t build anymore. He realized that to do so he needed to go the greenfield route. In a real sense the project rests on the premise that the Bay Area won’t liberalize its land-use policies. In the very long term, though, there’s no reason America shouldn’t have larger conurbations. California Forever envisions San Francisco and Sacramento someday being such a conurbation. California Forever would be at the heart of it. This can be positive-sum.
7/ Which goal seems more central to the project—building homes people can afford, or building a manufacturing hub?
Good question. My first instinct is the manufacturing. Sramek himself in our conversations seemed more animated by that aspect of it. But I approached them as an industrial policy researcher, so that may have biased the interactions. Having sat down with Gabe Metcalf, the urban planner, I’m confident that the housing and livability aspect is sincere too.
On sale everywhere The Conservative Futurist: How To Create the Sci-Fi World We Were Promised
