AI could be an extraordinarily fast decision-support system for finding an off-ramp that real estate developers like Donald Trump, Steve Witkoff, and Jared Kushner might miss.
The central unresolved issues include the Strait of Hormuz, Iran’s nuclear program, sanctions, and security guarantees. The June agreement demonstrated that Washington and Tehran can reach a framework, but competing interpretations—particularly over Hormuz—helped unravel it.
Here’s where AI could make a practical difference:
- Search millions of possible bargains. Feed an AI system the negotiating positions, sanctions, nuclear restrictions, shipping arrangements, inspection regimes, military deployments, previous agreements, and publicly stated red lines. It could generate thousands of packages that satisfy more of each side’s constraints than negotiators can manually explore.
- Find the smallest viable deal first. Don’t initially ask AI to solve the entire U.S.-Iran relationship. Ask: What combination of reciprocal actions could stop shooting and restore commercial passage through Hormuz for 60 days? The June MOU already provides a starting architecture: cessation of military operations, safe commercial passage and a defined negotiating period.
- Build an escalation simulator. Give policymakers a constantly updated model of: “If we strike X, what are Iran’s plausible responses?” It could examine missiles, drones, proxies, shipping, energy infrastructure, cyber activity and economic consequences simultaneously. Crucially, it should produce ranges and scenarios rather than pretend it can predict Iranian behavior.
- Create an AI red-team. Have separate models argue the problem from American, Iranian, Israeli, Saudi, Emirati, Qatari, Omani and Chinese perspectives. The purpose isn’t to decide who’s right; it’s to expose assumptions such as “Iran will obviously respond this way.” Those assumptions are exactly where military planning can go badly wrong.
- Solve the Hormuz puzzle operationally. AI can combine satellite imagery, ship movements, drone and missile threats, insurance data, oil flows and naval information to model different arrangements for safe passage. That’s particularly important because Hormuz remains effectively closed and the disruption is damaging Gulf energy production and trade.
- Draft a face-saving agreement. This is underrated. Wars are sometimes easier to enter than leave because leaders need to portray the outcome as consistent with their stated objectives. AI could generate alternative formulations in which Washington emphasizes nuclear restraints and freedom of navigation while Tehran emphasizes sovereignty, sanctions relief and non-aggression—without requiring either side to endorse the other’s narrative.
The most interesting possibility is an AI-assisted “minimum agreement.” Rather than attempting another grand settlement, negotiators could model a sequence such as: verified cessation of attacks → protected commercial shipping → reciprocal military stand-downs → limited sanctions relief → nuclear monitoring → longer negotiations. Each concession would occur only after a separately verifiable action by the other side.
There is already evidence that pieces of such an arrangement aren’t imaginary. Iran and Oman have been working on a framework concerning Hormuz, although Reuters reported on September 12 that it does not provide for immediate reopening and that substantial conditions remain.
And AI has one peculiar advantage over human negotiators: it doesn’t care who gets credit. It can search relentlessly for arrangements in which both governments can tell their domestic audiences that they didn’t capitulate.
The final decisions would still need to be made by Trump administration officials, Iranian officials, military commanders and diplomats. AI would be valuable precisely because it could enlarge the menu of choices available to those humans—not make the political decision for them.
