“AGI” rhetoric is leaving an imprint on geopolitics.
Those who believe artificial general intelligence is imminent are seeing signs of it everywhere—this growing faith has geopolitical consequences.
What’s happening? A new “general AI agent” created by Manus AI became so popular that spiking server usage temporarily crippled it. The software is Chinese-built, Singaporean-owned, and the chief scientist behind its development has claimed it could be “a glimpse into AGI.”
So what? Glimpse or not, progress is being made towards an ultimate goal for labs and their deep-pocketed investors; those that have poured billions of dollars into the dream of creating autonomous software that can perform at a superhuman level across many different domains and contexts—a reasonable definition of “AGI”.
Based on Claude’s Sonnet 3.5, a fine-tuned version of Qwen, and with a “low-profile team” in China “backed by Chinese investors and developers”, Manus AI claims to have built something along those lines. It has done so at a time when geopolitical competition and fear surrounding “AGI” are on the rise.
Now what? Skepticism is definitely warranted. Experts have raised questions about Manus AI’s model usage, technology stack, capabilities, and the effectiveness of its product. More broadly, the entire topic of “AGI” is shrouded in definitional challenges, competing incentives, mis-interpretations, and a lack of cross-comparable evidence.
These factors have not slowed the rhetorical hype-cycle; and along with driving both philosophical and financial speculation, claims about imminent “AGI” are also stoking debates at the geopolitical level.
The visible hand. Some experts claim that software is getting close to being “better than almost all humans at almost all tasks”, to borrow a phrase from Anthropic CEO, Dario Amodei. The narrow tasks, like screening job applications, or finding real estate deals in Manhattan, demonstrated by Manus AI, do not equate with the full perceptive and holistic function of a conscious human mind.
As Santa Fe Institute Professor Melanie Mitchell and others have argued, general intelligence extends beyond “subserving the maximisation of reward” in the way that current probabilistic model software does. Efficiency-seeking bureaucratic models built by corporations are just one way of developing alternative forms of intelligence.
The geopolitical impact. Whatever you call it, as companies make progress building software that can act on a user’s behalf to solve practical problems, there is a growing consensus that the downstream economic, military and intelligence benefits of sophisticated artificial intelligence technologies will be significant.
To the victor, the spoils. Whether they count as real “AGI” or are just a new phase in the evolution of software, perceptions are that “AGI” could be a national security and economic game-changer. This perception is feeding an competitive political and industrial dynamic reminiscent of the nuclear arms race from the middle of the last century.
“The government knows AGI is coming.” In this recent conversation between Ben Buchanan, the Biden White House’s top AI advisor, and Ezra Klein of the New York Times, Klein notes, with the kind of reverence that’s usually reserved for the converted, that “person after person” including leading researchers and government officials have approached him claiming that “AGI” is “about to happen.” Furthermore, he thinks “they’re right.”
Buchanan doesn’t endorse the idea of imminent “AGI” exactly; he says repeatedly during the interview that he is not a fan of the term. But, noting concerns about the potential of ever more advanced software to exacerbate nuclear, chemical, biological, and cyber risks, he defends efforts to “identify a chokepoint that can differentially slow [China] down” in its competition with the United States.
Buchanan also cites the October 7, 2022 semiconductor controls put in place by the Biden administration, which sought to limit China’s access to cutting-edge GPUs useful for training frontier models, as an example of action by the previous administration to shore up the US’ technological lead in this critical new technology.
Commerce’s Framework for Artificial Intelligence Diffusion (the “diffusion rule”) was a parting step by the Biden administration to further solidify this dominance. Trump is likely to carry on with this part of the Biden administration’s tech agenda. It would not be surprising if US security hawks revive the prospect of tightening restrictions on China’s access to semiconductors in the wake of the Manus AI rollout.
How is the race unfolding? The new administration is encouraging businesses to embrace the rhetoric, and the reality, of the race towards “AGI.” It has doubled down on AI “dominance” as a theme, and is “clearing a path for the United States to act decisively to retain global leadership in artificial intelligence” with a flurry of orders and policy augmentations at the start of the year. Chris Wright, Secretary of Energy, even declared that the US is at “the start of a new Manhattan Project.” for artificial intelligence capabilities.
Industry is leaning in. Google co-founder, Sergey Brin, is on record saying that “the final race to AGI is afoot”, and that Google can cross the line first, if it can “turbocharge” its efforts.
OpenAI CEO, Sam Altman, has characterized the $500 billion Stargate Project that Trump backed earlier this year as a step “critical on the path” to AGI.
Microsoft, a key cash and compute contributor to OpenAI, reportedly has an agreement to see “AGI” as a sufficiently profitable bit of software that generates $100 billion.
Notably, Microsoft CEO Satya Nadella has publicly taken a slightly different view, arguing that “self-claiming some AGI milestone” is “just nonsensical benchmark hacking” and that “the real benchmark is the world [economy] growing at 10%.”
Outside of the US. In Europe, leaders recognize that private-sector innovation is the main vector of progress towards more advanced software applications that could deliver serious economic dividends—whether it is branded as “AGI” or not.
At the recent Paris AI Action Summit, French President, Emmanuel Macron said “at the national and European scale, it is very clear that we have to resynchronize with the rest of the world” in order to remain competitive in terms of regulation and innovation.
Behind the curve. As we wrote last week, this resynchronization will not be simple. Europe must now grapple with the question of how to balance its current reliance on American providers, and a lack of domestic expertise to satisfy growing demand for cutting-edge capabilities, specifically for military and intelligence applications that could someday roll up to “AGI”-like systems.
“The whole AGI rhetoric is about creating God,” Arthur Mensch, CEO of Mistral AI—one of Europe’s few legitimate contenders in foundational model research—told the New York Times, dismissing the “very religious” tendencies driving such development, while acknowledging that a “European champion” is necessary to shape the global balance of technological power.
To race or not to race? Fanatical, faith-based in some cases, and highly charged with geopolitical concerns; the unresolved race towards “AGI” is having tangible effects on policy, and inter-governmental dynamics.
Releases like Manus AI’s “general AI agent” out of China may not end up moving the dial much in a technical sense; but others are surely on the horizon that will. The geopolitical clamor surrounding “AGI” will only grow louder and more fraught as a result.
What we’re reading:
A CSIS report on biases in AI application in foreign policy use cases.
Reports on the escalation of the US-Canada trade war.
A view from Robert Diab on what Jevons Paradox means for regulating AI from a sustainability perspective.
What we’re looking ahead to:
12 - 14 March: G7 Foreign Ministers' Meeting, Quebec.
14 - 16 May: B7 Summit, Ottawa.
4 June: AI+ Expo and Ash Carter Exchange in Washington, DC.
June (expected): The UN Internet Governance Forum.
15 - 17 June: G7 Leaders’ Summit in Kananaskis, Alberta.
24 - 25 June: North Atlantic Treaty Organization (NATO) Summit.
6 - 7 July 2025: Annual BRICS Summit, Rio de Janeiro, Brazil.
9 - 11 July 2025: AI for Good Global Summit.
9 - 23 Sep 2025: UN General Assembly (UNGA 80), New York.
V helpful note, thanks