After the Trump-Pope Clash:
The Case for Statesmen, Not Institutions

×

The Trump-Pope Leo XIV clash reveals a hard truth: no single institution — not the Vatican, not the UN — will broker a global AI treaty. But the Trump administration itself has already named the alternative: "the prudence and cooperation of statesmen." That phrase, spoken by OSTP Director Michael Kratsios at the UN Security Council, is almost verbatim from the most important memo most people have never heard of — Henry Stimson's 1945 proposal to President Truman. The question is whether a coalition of advisors will act on that logic before the window closes, as it did eighty years ago.

By Rufo Guerreschi — April 2026

What just happened — and what it means

Over the past 48 hours, Trump attacked Pope Leo XIV as "weak on crime" and "terrible for foreign policy," posted an AI-generated image of himself as Jesus, and doubled down at Joint Base Andrews.[1] The Pope responded from the papal plane that he has "no fear of the Trump administration."[2] The U.S. Conference of Catholic Bishops rebuked the President; even Marjorie Taylor Greene condemned the Christ-like image as "an Antichrist spirit."[3]

For those working to catalyze a US-China-led global AI treaty, the implications are immediate. The Vatican remains a vital moral voice on AI — from the Rome Call for AI Ethics to the doctrinal note Antiqua et Nova. But counting on a Pope-Vance-Trump pipeline as the primary channel to influence AI policy just became significantly harder. That channel is now relegated to moral guidance, not policy. The political climate around it has turned toxic. We need to adapt.[4]

The governance gap

Leading AI scientists have warned that advanced AI poses catastrophic risks — from loss of human control to concentration of power.[5] Nobel laureate Geoffrey Hinton puts the probability of AI-caused human extinction at 10–20% within thirty years.[6] In October 2025, 53% of Americans believed AI would likely "destroy humanity," and 63% thought humans would lose control.[7] By June 2025, 57% rated AI's societal risks as "high," and concern about AI ending the human race rose from 37% to 43% in just three months.[8] Yet no governance framework exists.

The Stimson precedent

In September 1945 — five weeks after Hiroshima — Secretary of War Henry Stimson presented Truman with a memo that shaped the nuclear age. US-Soviet relations, he argued, were "not merely connected with but as virtually dominated by" the atomic bomb.[9] Unless the Soviets were "voluntarily invited into the partnership upon a basis of cooperation and trust," the result would be "a secret armament race of a rather desperate character."[10] He urged Truman to approach the Soviets directly — not through the UN, not through a coalition of allies.

Even before Stimson, physicist Niels Bohr had urged Roosevelt in 1944 to approach the Soviets about shared control of atomic energy before the bomb was used — warning that "any temporary advantage, however great, may be outweighed by a perpetual menace to human security."[11] Roosevelt was sympathetic; Churchill rejected the idea. Roosevelt died before acting on it.

Stimson was not alone. Deputy Secretary of State Dean Acheson backed the bilateral approach and chaired the committee that produced the technical blueprint. Former Vice President Henry Wallace went further, publicly advocating direct engagement with Moscow.[12]

Truman faced a three-way choice. Option 1: no treaty — keep the nuclear monopoly. Secretary of State Byrnes and military hardliners favored this. Option 2: an allied-designed treaty — negotiate with Western partners, then present to the Soviets. Option 3: Stimson's approach — go directly to Moscow, bilaterally.

Truman chose Option 2. The Baruch Plan — based on an Acheson-Lilienthal Report largely written by Oppenheimer — was technically solid but politically dead on arrival because the Soviets had not been consulted. They counter-proposed a Gromyko Plan within six days. Deadlock ensued. The Cold War followed.[13]

As Zaidi and Dafoe concluded for Oxford's Centre for the Governance of AI: radical schemes for international governance can get widespread support, but that support is "tenuous and fleeting."[14] Stimson's bilateral-first approach was the right one. Truman chose otherwise. The window closed.

Three options for Trump

Trump faces a strikingly parallel choice.

Option 1: No treaty. An unfettered race to superintelligence. This is the dominant faction — led by David Sacks and shaped by Peter Thiel, who has argued the greater risk is a treaty creating a totalitarian world government.[15]

Option 2: Multilateral or allied-first. The G7, OECD, or UN path. Kratsios has explicitly rejected this: the US "totally rejects all efforts by international bodies to assert centralized control."[16]

Option 3: US-China bilateral. Stimson's logic, updated for 2026. Kratsios pointed to this when he said the way forward runs through "the prudence and cooperation of statesmen." A position surprisingly close to that of UN Secretary-General Gutiérres, who said in 2023: "Only member states can create it, not the Secretariat of the United Nations."[17] From opposite directions, the UN and the White House converge: the solution must come from the states that actually build AI.

Trump and Kratsios appear to have already chosen Option 3 — at least in principle. But they are hesitant to execute. There is no public trace of a deal-making process, no evidence that key AI powerbrokers have been brought on board, and no indication the architecture has been stress-tested.[18]

Trump's attack on the Pope this week has strengthened the Option 1 faction. By placing the world's most prominent voice of moral restraint on the "enemy" side, he has made it politically harder to argue for governance frameworks inside the West Wing.

Yet the logic of Option 3 has not changed. Vance, or potentially Secretary of State Rubio — both devout Catholics with the standing to reach the president — could play the role Stimson played: making the case for bilateral engagement not out of idealism, but out of strategic calculation.[19] Stimson was no dove. He oversaw the Manhattan Project. His argument was that American security required cooperation, because the alternative — an arms race — was worse.

But one official is not enough. Acheson and Wallace backed Stimson; it wasn't sufficient. A critical mass of advisors — from the security establishment, tech industry, and faith community — will need to form if Option 3 is to prevail over Option 1.[20]

Three paths

Which option prevails — and how well the process is designed — will determine which path humanity takes. The race toward artificial superintelligence makes middle outcomes unlikely.[21]

Path A — Loss of control. Option 1 prevails, or Option 3 comes too late. ASI emerges ungoverned. Leading scientists assign 10–20% probability to human extinction. A coin flip, at best.

Path B — Power grab. Option 1 succeeds for one side, or Option 3 produces a flawed treaty. A durable global oligarchy — AI-driven surveillance and power concentrated in a few firms and states. As Putin stated: "Whoever becomes the leader in this sphere will become the ruler of the world."

Path C — Human triumph. Option 3 pursued with a properly designed process. A bold treaty prevents uncontrolled ASI and reduces concentration of power, unleashing AI's benefits for all of humanity.

Path C can only happen if co-led by the US and China. China has repeatedly called for global AI governance. The question is whether the United States will lead — or be led.

The process matters as much as the will

Steve Bannon has called for an "immediate treaty." Glenn Beck co-signed the Pro-Human AI Declaration. Most AI lab leaders — Hassabis, Amodei, Musk, Altman — have called for strong global AI governance. And 77% of US voters favor creating a strong international agency to safeguard against AI's worst risks, including 71% of Republicans.[22]

What is missing is not the will. It is the process. A bilateral deal-making framework must succeed where the Baruch Plan failed: attracting bipartisan support, withstanding political headwinds, avoiding the traps of authoritarianism, and credibly engaging China.

Kratsios and others may have already designed such a process, awaiting the right political moment — a private channel with Xi, or a summit window later this year. Or perhaps not. We do not know. What we do know is that the design of the deal-making architecture is at least as important as the political will to initiate it.[23]

Our coalition has produced a 356-page Strategic Memo with 24 contributors — including former officials from the UN, NSA, and WEF — analyzing precisely this: how a US-China-led treaty-making process can be designed to succeed, what risks it must avoid, and how to attract the support of the key influencers who will determine whether it happens.[24] We are building toward a closed-door roundtable in Washington, DC — co-hosted with leading policy organizations — to stress-test the architecture and advance a coalition of support among key potential influencers of Trump's AI policy.[25]

The Deal of the Century

On June 14, 1946, Bernard Baruch presented history's boldest treaty proposal to the United Nations — calling for exclusive international control of all dangerous nuclear technologies. It was the most ambitious act of American diplomatic leadership of the twentieth century. It was also the day Donald John Trump was born.[26]

The Baruch Plan failed — not because its vision was wrong, but because the bilateral groundwork had not been laid. Eighty years later, the president born on the day of its presentation has an opportunity to succeed where Truman could not. By following the logic his own OSTP Director has already articulated — the prudence and cooperation of statesmen — Trump could initiate a bilateral process with China that would do for AI what the Baruch Plan aspired to do for nuclear energy. And because AI governance requires controlling technologies that also enable biological, chemical, and autonomous weapons, a successful AI treaty would eventually extend to all weapons of mass destruction — completing, at last, what Baruch proposed on the day Trump was born.

That would be, by any measure, the Deal of the Century.

Rufo Guerreschi is the founder and executive director of the Coalition for a Baruch Plan for AI, an international coalition of 10 NGOs and 40+ advisors working to catalyze a US-China-led global AI governance framework. The coalition's 356-page Strategic Memo details pathways to a historic global AI treaty.

NOTES

  1. Trump's 334-word Truth Social post, April 12, 2026, shortly after a 60 Minutes segment featuring Cardinals McElroy, Tobin, and Cupich. Full text at NPR, TIME, CNN. At Joint Base Andrews: "I'm not a big fan of Pope Leo." ↩
  2. Pope Leo XIV aboard the papal flight to Algiers, April 13, 2026. NBC News; AP/NPR. ↩
  3. Archbishop Coakley (USCCB): "Pope Leo is not his rival; nor is the Pope a politician." USCCB. Greene on X: "An Antichrist spirit." Newsweek. ↩
  4. On the Vatican's AI governance work: the Rome Call for AI Ethics (11 world religions, Microsoft, IBM); Antiqua et Nova (January 2025); Benanti's Coexistence Appeal signed by Bengio, Hinton, Russell, Harari. See CBPAI Strategic Memo, pp. 199–215. ↩
  5. CAIS Statement on AI Risk (May 2023), signed by Hinton, Bengio, Altman, Hassabis, Amodei, and hundreds more. See also Bostrom, Superintelligence (2014); Ord, The Precipice (2020). CBPAI Strategic Memo, pp. 19–40. ↩
  6. Hinton on BBC Radio 4's Today, December 27, 2024: "Not really, 10 to 20 [per cent]." The Guardian. AI Impacts 2022 survey (n=738): median 5–10% extinction probability. ↩
  7. Yahoo/YouGov, Oct. 23–27, 2025 (n=1,770). 53% "destroy humanity"; 63% "lose control"; 20% rated extinction "very likely." ↩
  8. Pew Research, June 2025 (n=5,023). 57% rated AI risks "high." YouGov, March–June 2025: concern rose from 37% to 43% (n=1,112). ↩
  9. Stimson, Memorandum to Truman, September 11, 1945. Office of the Historian; Atomic Archive. ↩
  10. Stimson, same memo. See CBPAI Strategic Memo, pp. 103–119. ↩
  11. Bohr, Memorandum to Roosevelt, July 1944. Also: Open Letter to the UN (1950). See Rhodes, The Making of the Atomic Bomb (1986), pp. 525–538; Bulletin. ↩
  12. Acheson chaired the committee producing the Acheson-Lilienthal blueprint. Wallace publicly advocated direct Soviet engagement. The opposing faction was led by Byrnes. See Zaidi & Dafoe (2021), pp. 15–22; Herken, The Winning Weapon (1980). CBPAI Strategic Memo, pp. 103–119. ↩
  13. On the Baruch Plan's failure: Zaidi & Dafoe (2021), pp. 28–35. The Acheson-Lilienthal Report was "written in large part by… Oppenheimer" (State Dept.); Bird & Sherwin, American Prometheus (2005). ↩
  14. Zaidi & Dafoe, PDF, p. 3. Allan Dafoe is now Head of AI Strategy at Google DeepMind. See also Roehrlich, Nature Reviews Physics (2025). ↩
  15. Thiel's March 2026 Angelicum lectures framed AI treaty advocates as abetting the "Antichrist." Sacks's AI policy described as "largely determined" by Thiel's vision. CBPAI Strategic Memo, pp. 81–85, 285–299. ↩
  16. Kratsios, UN Security Council, September 24, 2025. CSIS noted this marked "a sharper break" from prior US positions. Kratsios's AI framework (March 2026) has no international chapter. ↩
  17. Gutiérres, press conference, June 12, 2023. Also at Shanghai (July 2025): governing AI is "a defining test of international cooperation." ↩
  18. No public executive order or joint expert committee on US-China AI governance. Kratsios's Genesis Mission positions the US domestically; the international complement remains absent. CBPAI Strategic Memo, pp. 75–78, 140–152. ↩
  19. Rubio occupies the institutional position closest to Acheson's in 1945–46. Both Vance and Rubio are devout Catholics. See Zaidi & Dafoe (2021), pp. 15–22. CBPAI Strategic Memo, pp. 161–198. ↩
  20. The Pro-Human AI Declaration (March 2026) suggests a coalition forming outside the administration: Bannon, Beck, Rice, Nader, Bengio, Russell. CBPAI Strategic Memo, pp. 86–102. ↩
  21. On the three-way fork: CBPAI Strategic Memo, pp. 49–52. Bostrom, Superintelligence (2014), ch. 14–15; Ord, The Precipice (2020), ch. 6. ↩
  22. Program for Public Consultation, UMD, Feb. 2024 (n=3,610). 77% favor an international AI agency (R 71%, D 84%). By August 2025, 74% favored banning lethal autonomous weapons. ↩
  23. Zaidi & Dafoe (2021), pp. 28–35; CBPAI Strategic Memo, pp. 103–119. Experts calling for a Baruch Plan for AI: Hogarth (2018); Clark/Anthropic (2023); Bengio (blog, 2024). See endorsements. ↩
  24. CBPAI Strategic Memo (v2.6, 356 pp., 24 contributors): AI predicament (pp. 49–67); treaty-making process (pp. 103–119); enforcement (pp. 131–152); 12 influencer profiles (pp. 153–300); coalition strategy (pp. 86–102). ↩
  25. CBPAI's Deal of the Century initiative. DC roundtable date TBD (set jointly with co-hosts). See roadmap. The initiative targets the political bottleneck: no other organization systematically works to influence Trump's actual advisors toward treaty-making. ↩
  26. The Baruch Plan was presented June 14, 1946. Trump was born June 14, 1946. Office of the Historian. On AI governance subsuming WMD governance: CBPAI Strategic Memo, pp. 131–152. ↩

cbpai.org · LinkedIn · X