Two senior US lawmakers have written to President Donald Trump, urging him to press Chinese leader Xi Jinping to commit that humans — not artificial intelligence — must retain control over all nuclear weapons launch decisions. The letter, dated 15 September 2025, comes ahead of a planned Trump-Xi meeting next week and reflects growing bipartisan anxiety over AI's role in nuclear command-and-control systems.

The Core Demand

Representative Ted Lieu, a California Democrat, and Senator Ed Markey, a Massachusetts Democrat, co-authored the letter addressed to President Trump. Their central ask: obtain from Beijing a direct commitment to preserve what they describe as the 'human-in-the-loop' principle in the deployment of nuclear weapons.

'AI should never launch a nuclear weapon by itself,' the lawmakers wrote, framing the issue as both a strategic and existential imperative.

The US Legal Basis

Lieu and Markey cited Section 1638 of the Fiscal Year 2025 National Defence Authorisation Act as the foundation of existing US policy. According to their letter, the provision explicitly declares that AI must not compromise the integrity of nuclear safeguards or remove human decision-makers from the employment of nuclear weapons.

'The United States already requires that the use of AI should not compromise the integrity of nuclear safeguards or compromise the principle of maintaining a human-in-the-loop in the employment of nuclear weapons,' they wrote. Lieu's office noted that both lawmakers authored the provision themselves, making their appeal to Trump a direct extension of their own legislative record.

What They Are — and Are Not — Saying About China

Notably, the letter stops well short of accusing Beijing of having authorised AI to independently trigger a nuclear launch. It provides no details about China's existing nuclear command architecture or how AI may currently be integrated into it. The request is, in essence, a pre-emptive diplomatic push — seeking a Chinese commitment to a principle the US claims to already follow.

'We urge you to get General Secretary Xi to commit the People's Republic of China to the same principle that the United States already follows,' the two lawmakers wrote. 'Since the United States follows this basic principle, so should the PRC.'

The Risk They Want Trump to Address

The lawmakers framed direct presidential engagement as critical to preventing a nuclear disaster — whether triggered deliberately or by accident. 'Your direct engagement on this matter is critical to ensuring AI doesn't accidentally or intentionally cause a preventable nuclear disaster,' they wrote.

This comes amid a broader global debate over the militarisation of AI, with several defence analysts warning that autonomous systems in high-stakes domains introduce unpredictable escalation risks. The Trump-Xi meeting, if it proceeds, could offer a rare opening to establish even an informal understanding on AI and nuclear safety — a domain where no binding international framework currently exists.

What Happens Next

It remains unclear whether the White House will formally include AI nuclear safeguards on the agenda for the planned meeting. As of the publication of this report, neither the Trump administration nor Beijing had publicly responded to the lawmakers' request.