Open Letter to the Presidents of China and the United States
(Version of December 4th, 2024)
[This Open Letter to the Presidents of China and the United States to join the Coalition for a Baruch Plan for AI will be regularly updated and, starting January 2025, periodically submitted to the Presidents of China and the US - and to their relevant special advisors and security agencies, their Ministries of Foreign Affairs and Missions to the UN in Geneva and New York - together with the latest version of our Case for a Coalition for a Baruch Plan for AI (pdf), and with supporting country-specific documentation]
__________________________
Your Excellencies President-Elect Donald Trump and President Xi Jinping,
In this letter, we urge you, your security agencies and your leading AI firms to engage in a global treaty-making process and engineering effort for AI of radically unprecedented scope, urgency, and effectiveness - open to all states on equal terms.
The escalating and unchecked advances of AI and the ongoing winner-take-all AI arms race between your two countries pose immense and urgent risks for each of your countries to be subjugated by the other, or for AI to result in immense human safety harm or even human extinction, due to accidents, misuse, proliferation or loss of human control.
You have an immense responsibility, together, to protect your citizens, humanity and future generation from immense harm as well as an enormous opportunity, together, to leave a positive legacy far greater than any other leader in the history of humanity.
As an expanding coalition of NGOs and experts from around the world with deep multidisciplinary expertise, our Coalition for a Baruch Plan for AI urges you to consider entering into an agreement along these lines:
__________________________
Memorandum of Understanding between China and the US on the Global Governance of Artificial Intelligence
[Date]
Our countries recognize that Humanity stands on the brink of an irreversible slide towards AIs capable of subjugating or annihilating humanity or enabling a single state or entity to do so.
Leading Chinese and US AI scientists have warned that catastrophic safety risks could materialize "at any time", while an AI-driven concentration of power and wealth rapidly unfolds, accruing immense unaccountable power in the hands of firms, terrorists, and malicious entities.
While nearly all states stand completely powerless, on their own, being forced into a reckless, winner-take-all AI race-to-the-brink, along with our leading firms and allies, each in fear of being subjugated by the other.
We still have a chance to prevail in this greatest challenge humanity has ever faced and realize the astounding positive potential of AI, though only through unprecedented global coordination. Preventing AI proliferation and unbridled advancement requires wide global compliance and collaboration and cannot be achieved via surveillance alone, as has been achieved so far from other dangerous technologies.
We must proceed with the utmost urgency and care to create global organizations that reliably prevent the immense and urgent risks of AI. These risks include loss of human control, concentration of power in a single state, catastrophic misuse and accidents, or an AI-driven global nuclear conflict.
The scope of the risks, the impossibility of two or a handful of states to succeed on their own, and the utter failure of predominant treaty-making models require the adoption of a radical yet time-proven treaty-making method. This method should be efficient, extremely authoritative, and time-bound by set dates. It should also be sufficiently inclusive and impervious to veto deadlocks and delaying tactics.
Our two countries commit to dedicating the highest communications and diplomatic effort and financial resources to pursue a global treaty-making and engineering initiative for AI. This initiative will have unprecedented scope, urgency, and effectiveness - with the full-time involvement of thousands of diplomats, national security officials, leading firms,s and independent experts. It will involve the following:
(A) Foster an urgent temporary US-China treaty on AI to define and implement—with the utmost urgency—extreme, dual-use, temporary, stop-gap measures to protect against extreme AI Safety risks—even at the cost of a substantial, global, extended slowdown or even pause in AI capabilities advancement.
(B) Advance, co-design, and enact with the utmost urgency a structured treaty-making process leading up to an open, global intergovernmental constituent assembly for AI, guided by these minimum terms:
1) A minimum and maximum mandate for such an assembly is set to ensure that the resulting global organization is sufficient in scale, delegated functions, and resources to reliably manage the risks in the years and decades to come, inspired by the 1946 Baruch Plan, while maximizing an overall increase in states' and citizens' sovereignty, and private initiative.
2) A set start and end date, ensuring at least two months of intense on-site on-site negotiations among thousands of high-level political, diplomatic, and security officials. Decreasing super majority rules should be adopted to ensure near unanimity., No state should have a veto, and timelines are obligatory. The start date will be set to six months after at least 60 states, accruing 50% of the world population and GDP, and three members of the UN Security Council have joined.
3) Rules for states’ vote weighting largely based on GDP and population, in consideration of the fact that more than three billion people are unconnected, illiterate, or unaware of the enormous asymmetries of power, wealth, AI safety expertise, and geopolitical realism.
4) A Global Public Benefit AI Lab and ecosystem will be urgently created with a cumulative investment of at least several hundred billion dollars to develop and jointly exploit the most advanced safe AI and enable states and firms a safe and profitable exit ramp from the ongoing uncontrolled AI arms race. It will be democratically governed, largely decentralized, public-private, and open to all states. It will be a consortium similar to a global AirBus for AI. It will be much better funded than any other AI Lab, primarily via sovereign funds and pre-commercial procurement contracts of participating states.
By coming together with intent, like never before, we can shape a safe and equitable AI future and affirm a global governance model for other dangerous technologies, our global digital sphere, and other global challenges for generations to come.
For the United States of America:
[Signature]
[Title]
[Date]
For the People's Republic of China:
[Signature]
[Title]
[Date]
_____________________________________________
For more information on the Coalition for a Baruch Plan for AI, please refer to our 2,000-word Executive Summary. A 60-page Case for a Coalition for a Baruch Plan for AI is also available on request.
Reach out at cbpai@trustlesscomputing.org