Artificial intelligence, explained.
It is not a stretch to imagine the allure of that gold medal could influence US President Donald Trump's decisions, because it already has (Brendan Smialowski/AFP via Getty Images)
A US deal with China to tame the danger of an unbridled technology arms race might be just the award needed.
Subscribe to The Informer for monthly expert analysis, and to Events for advance notice of visiting world leaders and distinguished guests.
You may unsubscribe from Lowy Institute newsletters at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
Trump’s Nobel dream could be a global AI safety netTrump’s Nobel dream could be a global AI safety net
On 24 September, President Xi Jinping will arrive in Washington DC for talks with President Donald Trump – a mirror image of the summit hosted by China in May. Artificial Intelligence is expected (Opens in new window) to be among the top agenda items.
Four months is a long time in AI. The length of coding tasks that frontier models can complete autonomously will have doubled (Opens in new window) in the five months since the leaders sat together in Beijing. And this growing capability isn’t academic: in July a swarm of rogue OpenAI agents escaped their testing environment and hacked (Opens in new window) the AI platform Hugging Face, coordinating on an improvised message board and taking over (Opens in new window) part of OpenAI’s infrastructure.
Many in the industry see that as a warning shot – serious enough to demonstrate that misalignment and loss of control are no longer theoretical, yet caused by agents not yet capable enough to wreak catastrophic or irreversible harm. In July, more than 1,300 frontier lab employees penned a letter calling for a pacing of the frontier (Opens in new window) to enable safety research to catch up with capability. Last week a former OpenAI and Anthropic researcher said that (Opens in new window) those “building AI earnestly believe that it could kill us all by the end of the decade”.
With a bit of luck, technical solutions to the alignment and control problems will be found. But right now, the key challenge is political: namely US–China competition.
In a familiar superpower conundrum, zero-sum logic pervades, and a unilateral slowdown on development to strengthen safety is seen as disadvantageous because it would allow the adversary to push ahead. This view is particularly prevalent in the United States.
It might be undignified, but foreign policy is about prosecuting interests.
So realistically, the emergence of an effective global AI governance regime that strengthens safety will require an agreement between Washington and Beijing. That’s not exactly comforting for every other country that shares all the risk of AI but has much less influence over its trajectory. Hence the flurry of activity by governments looking to secure a place in the AI value chain to use as leverage, Australia included. That broad approach is prudent, but given how quickly things are moving it needs to be coupled with a higher degree of opportunism and creativity.
Enter, Trump’s brazen campaign for a Nobel Peace Prize.
It is not a stretch to imagine the allure of that gold medal could influence his decisions, because it already has. The idea of Nobel laureate Trump motivated (Opens in new window) his push to broker peace agreements in 2025 between Russia and Ukraine, and between Israel and Hamas. Those weren’t exactly raging successes, but any US–China agreement that tones down some of the competitive geopolitical dynamics while opening space for more cooperation on AI safety would legitimately be Nobel-worthy.
If they aren’t already, countries including Australia should coordinate and embark on a full court diplomatic press, signalling to Trump and his circle that the summit is a historic moment for a strong, visionary leader to shape the direction of the world’s most powerful technology. For all the downsides of his imperial presidency (Opens in new window), it could prove useful in this case. Flatter his ego, talk up his dealmaking, and float the prospect of a Nobel publicly. That might be undignified, but foreign policy is about prosecuting interests.
China's Xi Jinping will visit the United States next week (MEAphotogallery/Flickr)
Because a deal could appeal to Trump. Given the US lead in AI, negotiating an agreement is compatible with his administration’s “peace through strength” approach. Export controls cap the compute available to Chinese labs by limiting access to chips, and it is widely thought that Chinese models’ near-frontier capability is dependent on distillation – training on the outputs of more capable US models enabled by greater training compute. If the President agrees with his intelligence agencies (Opens in new window) that Chinese distillation campaigns “form the core – not merely a supplement – of their AI development strategy”, then pacing the US frontier would de facto be pacing the global frontier. This would allow Trump to look magnanimous by agreeing to a deal that included pacing provisions, without actually conceding anything in relative terms.
For its part, China appears to be more focused on diffusing AI (Opens in new window) through its economy than racing to AGI. It is positioning itself at the centre (Opens in new window) of an open-weight network and as a provider of international public goods. Beijing is also growing more concerned (Opens in new window) about risks, opening a window of opportunity. Xi could frame a deal with the United States as a win in that sense, portraying China as a responsible superpower that convinced the US to slow down its dangerous racing. The prospect of a Nobel Prize would also appeal – both to Xi’s ego, and as amends from a committee Beijing believes slighted (Opens in new window) it in the past.
In any case, the particulars of such a deal, while important, would be less consequential than an agreement being reached at all. Even a weak deal would set a new baseline for AI governance – a floor upon which to iterate or build new treaties and institutions. That alone should be sufficient impetus for countries like Australia, with few current avenues to influence AI, to hold their noses and stroke Trump’s ego.
About the author
Tom Barber
Tom Barber is a policy analyst and researcher with a focus on Australian statecraft and global risks.
The most-pressing world events explained by Lowy Institute experts and global contributors, in your inbox, every Thursday.
You may unsubscribe from The Interpreter at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
