Geneva MOU's Hardest Test Is Verification, Not Signatures

The Geneva memorandum delivered something the past year could not: signatures on the same page from a sitting US vice president and the speaker of Iran’s parliament. That was the easy part. The agreement, signed by JD Vance and Mohammad Bagher Qalibaf, is a framework. Whether it survives the summer depends on a thinner, less visible track that the public will see only in headline form: verification.

Frameworks fail on verification. The 2015 JCPOA shipped with a verification annex that became the document’s most-litigated section, and the deal collapsed not over signatures but over what could be inspected, when, and how. The Geneva MOU enters that same gauntlet now, with three differences worth noting up front.

First, the International Atomic Energy Agency has not signed. Vienna has kept public silence since the framework was announced, a deliberate posture from a body that needs to be invited into a verification track rather than be seen leading it. The Board of Governors meets on a fixed quarterly schedule; an intersessional statement would be a signal that the IAEA has been briefed on what to monitor. The absence of one suggests the access architecture is still being negotiated, not implemented.

Second, the framework was reached under active fire. Israeli operations into Lebanon continued into the second day after the Geneva text was initialed, and Tehran’s response to Washington’s escalation rhetoric remains, by official channels, silence. A verification system built while strikes are ongoing is structurally different from one built in a freeze. Inspectors do not deploy into contested airspace, and chain-of-custody on environmental samples requires logistics that war complicates.

Third, the political cover on both sides is thin. Qalibaf can sign as parliament speaker, but the Supreme Leader’s office can disown the text without retracting the signature. The US Senate has not voted on anything. The MOU is binding on the executives who signed it for as long as those executives choose to be bound — a fragile foundation for the multi-year monitoring cycles a real nuclear agreement requires.

What verification will actually look like

A working verification system, if it emerges from the Geneva track, will be built around four elements.

Declared site access: routine inspections of facilities Iran has already acknowledged, primarily Natanz, Fordow, and the heavy-water plant at Arak. The procedural questions here are about frequency, notice windows, and inspector nationality, not about whether access happens at all.

Undeclared site access: the harder category, where the IAEA must be able to follow procurement and intelligence leads to facilities Iran has not declared. The 2015 deal called this “managed access” and built in a 24-day clock that critics argued was long enough to scrub evidence. Any successor language will be litigated against that benchmark.

Sampling protocols: environmental swipes, centrifuge component accounting, and uranium hexafluoride flow tracking. These are the technical guts of the inspection regime, and the agreements that govern them are usually classified annexes rather than public text.

Data continuity: a guarantee that monitoring cameras and seals operate without gaps. This is the issue that broke the 2022–2024 cycle, when Tehran removed surveillance equipment and the agency lost its baseline. Rebuilding continuity from a cold start is a months-long process, not a same-week one.

None of these will be visible to a reader watching the wires. What will be visible is the absence of friction — quiet weeks, IAEA statements that do not flag anomalies, no Iranian announcements of new cascades. Verification, when it works, is boring.

Snapback as the back-end test

The 2015 framework included a snapback mechanism that allowed any party to trigger reimposition of UN sanctions. That clock ran out in October 2025. Any new agreement will need its own enforcement architecture, and the Geneva MOU language on that point has not been released publicly.

The question is not whether snapback exists but who controls the trigger and what counts as a triggering event. A maximalist US position would treat any IAEA-flagged anomaly as cause; a maximalist Iranian position would require UN Security Council action with the attendant veto math. The drafting choices between those poles are where the deal is being made or unmade right now, behind closed doors in Geneva and on classified channels in Vienna.

Near-term signals worth tracking

The Friday window matters. President Trump’s pledge on Hormuz transit sets a public deadline that doubles as a verification cue: a smooth reopening would indicate the Geneva track is being honored on the maritime file, while continued interdictions would suggest the framework is not yet operational below the headline level. Tanker insurance rates and Lloyd’s List notices will move before official statements do.

The Vienna calendar is the second cue. An out-of-cycle IAEA Board statement, even a brief technical one, would mark the verification track moving from negotiation to setup. Continued silence past the end of the week would suggest the access architecture has not yet been settled.

The third cue is Tehran’s domestic press. Outlets aligned with the Supreme Leader’s office have not yet commented on the MOU’s substance. Sustained silence reads as permission; a critical editorial would be a signal that the framework is being repudiated from inside the system that has to live with it.

The Geneva MOU bought time. What it has not yet bought is a verification system, and time without verification is the condition under which every prior Iran framework has died. The next two weeks will determine whether this one is different.