Biden and Xi agree to keep AI away from nuclear weapons
President Joe Biden and Chinese leader Xi Jinping just reached a landmark agreement to bar artificial intelligence from controlling nuclear weapon systems. Meeting at the APEC summit in Lima, both leaders committed to keeping human control over any decision to deploy nukes. Why it matters: It establishes a rare and vital military baseline before a new US administration takes over.
Hollywood's darkest sci-fi trope usually involves an autonomous supercomputer taking over the nuclear football and deciding humanity's fate in milliseconds.
In the real world, the two biggest global superpowers just agreed that nightmare should stay in the movies.
Meeting for just under two hours on the sidelines of the APEC summit in Lima, Peru, President Joe Biden and Chinese leader Xi Jinping reached a landmark breakthrough: both nations agreed that artificial intelligence must never be given control over nuclear weapons systems.
White House National Security Adviser Jake Sullivan told reporters after the meeting that the two sides committed to ensuring "there should be human control over the decision to use nuclear weapons."
An unexpected guardrail
The agreement is particularly striking given how frosty bilateral military talks have been.
Over the past four years, the Biden administration pushed repeatedly for guardrails around nuclear safety and advanced technology, only to be rebuffed by Beijing. In July, China officially froze working group meetings on nuclear arms control in protest of American weapons sales to Taiwan.
Pulling off a concrete consensus on military AI in the final weeks of Biden’s presidency came as a genuine surprise to diplomatic observers.
Beyond AI, Biden also pressed Xi to rein in North Korea's military support for Russia in Ukraine, and signaled headway toward securing the release of two American citizens detained in China. But keeping autonomous code away from the nuclear trigger was the clear centerpiece.
Why it matters
We think this agreement represents a rare, crucial pause in what has otherwise felt like a reckless race toward militarized autonomy.
As militaries worldwide scramble to plug large models and algorithmic decision-making into target acquisition, sensor networks, and drone fleets, the risk of rapid escalations has soared. If automated logic or an algorithmic hallucination were ever allowed to feed directly into strategic launch systems, the buffer time for human sanity drops to absolute zero. Locking in human-in-the-loop oversight at the apex is the absolute bare minimum for planetary survival.
That said, the deal has obvious blind spots.
For one, it is a statement of diplomatic intent, not an arms control treaty with real verification mechanisms. There is no technical blueprint to audit whether a military command center is quietly running automated heuristics behind closed doors.
More importantly, with Donald Trump preparing to re-enter the White House, it remains to be seen whether the incoming administration will honor or build upon Biden’s eleventh-hour diplomatic pacts. For now, it is a smart and necessary red line on paper, but keeping that red line intact will be the real test.
Sources (1)
- politico.com(opens in a new tab)Originalpolitico.com