The hardest party to manage in the Iran talks isn’t Iran

Sometime this spring, at the height of Washington’s most delicate Middle East negotiations in a generation, American officials did something for which the US-Israel relationship offers scarcely any precedent: they quietly asked other governments to warn Iran of a possible Israeli plot to assassinate Tehran’s two chief negotiators.

That is the substance of a New York Times report published earlier this month; two US officials have confirmed the warnings to CNN, while Israel has dismissed the report as a fabrication. Washington feared Israel was plotting to kill Iranian Foreign Minister Abbas Araghchi and Iranian Parliament Speaker Mohammad Bagher Ghalibaf, the men leading Iran’s side of the talks. Unable to order its ally to stand down, Washington warned its adversary about its friend. Whether or not a plot existed, the decisive fact is the conduct: Washington judged the danger real enough to act on, and acted.

In this phase, Washington’s most difficult task is not simply keeping Iran at the table; it is preventing its closest ally from removing the table altogether. The hardest party to manage is not the one Washington spent two decades treating as an implacable enemy. It is the one it arms.

A pattern, not a hypothetical

According to the Times, Israeli strikes earlier in the war killed Ali Larijani, then secretary of Iran’s Supreme National Security Council, and Kamal Kharazi, a former foreign minister and foreign policy adviser to the supreme leader—both pragmatic figures involved in the talks and people Washington had hoped to negotiate with. The channel today runs through Araghchi and Ghalibaf partly because the men who might otherwise have led it are dead.

Ghalibaf himself has reportedly survived two Israeli assassination attempts, one in the 12-day war of June 2025 and one this year, when Israel struck a bunker where senior officials were meeting.

The spoiler playbook fails when the spoiler is a friend

Conflict-resolution scholarship calls actors who see a peace process as a threat and act to destroy it “spoilers”. Political scientist Stephen Stedman’s foundational work observed that spoilers outside a process are the more dangerous kind, bearing no cost when talks collapse and gaining what they want when they do. The literature is also precise about timing: spoilers strike when a process nears real achievement, or when a charged symbolic moment can turn an incident into a rupture.

By that standard, the funeral period created almost textbook conditions for spoiling. The US-Iran track had just produced an interim agreement to halt the war and reopen the Strait of Hormuz. At the same time, Iran was holding days of public mourning for the late Supreme Leader Ayatollah Ali Khamenei, killed on the war’s first day, with the processions overlapping with US Independence Day. Israeli Defence Minister Israel Katz had declared Iran’s new supreme leader, Mojtaba Khamenei, marked for death, while Araghchi had promised a forceful response and demanded that Washington restrain its ally. So acute was the danger that Mojtaba Khamenei stayed away from the public funeral ceremonies for his own father. A diplomatic process nearing a breakthrough had collided with a moment of maximum political and symbolic exposure: precisely the kind of opening the spoiler literature warns about.

Everything the literature prescribes for managing spoilers, including inducement, socialisation and coercion, was designed for adversaries. None of it fits the distinctive and under-examined configuration Washington now faces, in which the suspected spoiler is the ally at the centre of its regional strategy. Coercing Israel is politically unthinkable in Washington. Inducing it is redundant; it already receives the full package. Socialising it into a process that its officials regard as a strategic disaster because it forecloses regime change and releases funds to Tehran is a contradiction in terms.

Israeli reporting explains why. An investigation recently published by Israeli news site Ynet documented how Israeli Prime Minister Benjamin Netanyahu’s office pressured its own intelligence establishment to certify inflated assessments of the war’s achievements, over objections from officers and scientists; the investigation described the agreement as a very bad one for Israel. If the war’s declared achievements outran its actual ones, a durable agreement is not merely unwelcome to Israel; it is narratively dangerous, because every month the process survives is an audit of the victory Israelis were told they won.

Israel is not the only spoiler pressing on the process; Iran’s system has produced internal ones. Days after the memorandum was signed, strikes on Gulf targets continued, and the Islamic Revolutionary Guard Corps (IRGC) threatened to abandon the talks, even as the government that signed it defended the negotiating process. Scholars of the region’s peace processes, notably political scientist Wendy Pearlman, have shown that spoiling is often the work of factions contesting their own camp’s leadership. The process is squeezed from two directions: an allied spoiler outside one principal, and factional spoilers inside the other. Any workable definition of stabilisation must cover both.

Alliance theory has a name for Washington’s predicament. Since political scientist Glenn Snyder’s classic work on the alliance security dilemma, scholars have called it entrapment: a patron dragged by a client into outcomes it does not want. In the standard version, the ally drags you into a war. The Iran file presents the inverted case, a client working to drag its patron out of a peace. Snyder’s framework also explains why: an ally with enough standing in the patron’s domestic politics to resist coercion, facing a patron that cannot afford to let it defect, holds bargaining power out of proportion to its dependence.

When the spoiler is an ally, spoiler management collapses into alliance management, and that is not going well. After Israeli ministers attacked the memorandum, US Vice President JD Vance told reporters that if he sat in Israel’s cabinet, he “might not be attacking the only powerful ally” it has left. The sequence reads as a ladder: a private request that Israel refrain, then warnings routed through intermediaries, then a public rebuke from the vice president, then a leak to the press. Each rung is more public than the last, suggesting that Washington did not consider the one below it sufficient. Through Snyder’s lens, the ladder is not tactical clumsiness but structurally inevitable: the only path open to a patron that can neither coerce nor abandon. The next rung may be climbed in person: US President Donald Trump and Netanyahu agreed in a July 3 call to meet soon in the US. A superpower reduced to this is conceding that it does not control its partner. That concession is the central fact of this negotiation. It points to the need for a fourth tool the spoiler literature has yet to catalogue, stabilisation through third parties, protecting the process itself until the moment of danger passes.

Measure this phase by what doesn’t happen

US officials acknowledged that once serious negotiations began, the calculus of targeting Araghchi and Ghalibaf inverted: killing them would kill the talks. Trump made the calculus explicit: Iran’s leaders were gathered in one place, but striking would leave him with “nobody to negotiate with”.

The intervals between rounds are where peace processes die: the table imposes discipline and the days between rounds do not. The funeral and the Fourth of July pushed domestic rhetoric to maximum volume just as the channel was most exposed.

In such a phase, progress is the wrong objective; the right one is stabilisation: freezing the situation, tacitly and through intermediaries if necessary, until the symbolic moment passes. The weeklong pause in talks that both sides accepted for the funeral reflected this logic. It did not, however, stabilise the wider conflict. US strikes resumed on July 7, before the funeral ceremonies had ended, and Iran responded with attacks on US facilities in the region.

The fourth tool is not hypothetical; the Times shows it in operation. When Iran feared its delegation would be attacked en route to talks in Islamabad, it sought guarantees through Pakistani and Qatari intermediaries that its negotiators would not be targeted. Pakistani fighter jets escorted the planes to the talks and back. When a threat nevertheless emerged, forcing the delegation’s aircraft into an emergency landing, the team drove eight hours home and later continued negotiating in Doha and Switzerland. Third parties are absorbing risk the principals cannot carry: vouching for what the patron cannot guarantee, saying to Tehran what Washington cannot say aloud. Agreements rarely die at the table; they die between rounds, when a charged moment meets an actor who wants them dead, and no one has built the machinery to absorb the blow.

The memorandum will ultimately be judged on Hormuz, centrifuges and sanctions relief. For now, the achievement is narrower: despite the renewed US-Iran strikes, the feared Israeli operation did not take place, the negotiators remained alive, and the diplomatic channel stayed open. That was success of a limited kind, earned the hard way: not by managing an enemy, but by managing a friend.