Diplomacy on a deadline

Many thought leaders have argued we may eventually need some international agreement on AGI. For example, Sam Altman has repeatedly argued we may need an “IAEA for AI”. More recently, a group of 300+ prominent figures called for an international agreement on AI red lines by 2026. In this blog, we have explored a range of international cooperation models from managed proliferation, to a CERN for AI, to a tax avoidance treaty, to a Cosmic Endowment Fund.

Yet, negotiations for international agreements can take a lot of time. For example, at the UN, diplomats have been debating the definition of “terrorism” since 1996 for the Comprehensive Convention on International Terrorism. The country experts meet once a year, they agree to disagree, they meet again next year. Even functional treaties can take a long time to negotiate. The Chemical Weapons Convention is a success story. It has resulted in the destruction of 99%+ of the world’s declared chemical weapons stockpile for which it received the Nobel Peace Prize. Still, negotiations started in 1980, it was adopted in 1992, and entered into force 1997.

At that speed, even if diplomats were to start negotiating an international agreement for AI today, it would not enter into force before 2042. In AI time, that’s an eternity. Indeed, speed is not just an issue for any international agreement on AI itself, but also for managing potential future issues in non-AI-domains that might emerge if AI manages to accelerate science and thereby increase the rate of societal, economic, environmental, and technological change.

The good news is that there is no physical law that states that international agreements have to be slow. So, I decided to explore examples of fast and ambitious international agreements. The case studies cover a range of contexts. Still, based on reading inside accounts of them (Camp David, Dayton, Reunification, Montreal, Smallpox), some common patterns emerged: From informal connections and processes, to breaking down problems into manageable chunks, to engineering time pressure.

Case studies of fast minilateral agreements

1. Camp David Accords

Egypt and Israel fought five wars (1948, 1956, 1967, 1967-1970, 1973). The Accords in 1978 created a permanently normalized relationship between Israel and its largest Arab neighbor in exchange for the return of Sinai. There was no pre-negotiated text. Except for US President Jimmy Carter, everyone from his own staff to the Egyptian and Israeli delegations expected the Camp David talks to fail. In 13 days, Carter, the Egyptian president Anwar Sadat, and the Israeli prime minister Menachem Begin, delivered one of the most significant agreements in the region’s history.

2. Dayton Agreement

The break-up of former Yugoslavia along ethnic lines led to a series of bloody conflicts including the Bosnian War and the Croatian War of Independence. The US followed a “talk, talk, bomb, bomb” strategy, enforcing a no-fly-zone, and eventually even conducting air strikes against the Bosnian Serbs. In 1995 the US negotiating team brought the presidents of Serbia (Milosevic), Bosnia (Izetbegović), and Croatia (Tuđman) to the Wright-Patterson Air Force Base in Dayton, Ohio where they managed to reach an agreement within 21 days. The agreement ended two wars and resulted in a peaceful transfer of Slavonia to Croatia as well as the establishment of Bosnia and Herzegovina as a new country with autonomy for the region controlled by ethnic Serbs.

3. German Reunification

On November 9, 1989 the Berlin Wall fell more or less to the surprise of everyone. Within less than a year, by October 3, 1990, the formal Reunification of Eastern and Western Germany took place. In this process the West German Chancellor Helmut Kohl did not only have to contend with his coalition partner FDP, the political opposition of the SPD, and the East German people and government. Kohl had to get the sign off from Germany’s four post-WW2 occupying powers: the United States, the Soviet Union, the United Kingdom and France. The latter three were initially hesitant about German Reunification and the Soviet Union alone had ca. 400’000 troops stationed in Eastern Germany in 1990 (that’s more than the initial Russian invasion force of Ukraine in 2022). German Reunification and the Treaty on the Final Settlement with Respect to Germany provided a significant gain in German sovereignty and had a lasting impact on European Integration, and on the European security architecture.

4. US-EU Trade Deal

Trade deals are slow, drawn-out affairs. On average it takes 1.5 years of negotiations for the US to sign an agreement and over 3.5 years to reach the implementation. EU trade deals tend to be even slower. The negotiations for a Canada-EU trade deal took 7 years to reach signature. Negotiations on an EU-Mercosur trade deal started in 1999. It was signed in 2019, and has still not been implemented. A EU-UK agreement was reached within 10 months under the looming deadline of Brexit. The recent trade deal between the US and the EU was negotiated in only 4 months from “Liberation Day” on April 2 to July 2025. While many parties in Europe are unhappy with this deal, it does show how much an artificially created crisis with an artificial deadline can accelerate negotiations.

Case studies of fast multilateral agreements

5. Montreal Protocol

Chlorofluorocarbons (CFCs) are chemicals that are useful in a range of circumstances from refrigerators, to air conditioners, to hair sprays. Unfortunately, throughout the 1970s and 1980s evidence emerged that they damage the Earth’s natural Ozone layer which protects us all from UV radiation. The Ozone hole above Antarctica was first reported by British scientists in May 1985. Formal negotiations on a treaty to limit global CFC consumption began in December 1986. In September 1987 the Montreal Protocol on Substances That Deplete the Ozone Layer was signed. The treaty and its amendments have phased out about 99% of Ozone-depleting substances and saved millions of lives from skin cancer. Despite its focus on deregulation and skepticism towards internationalism, it was the US government under Ronald Reagan that pushed for this treaty with domestic industry support against resistance from the UK, France, Italy and developing countries.

6. Smallpox Eradication

Smallpox is a disease that has killed about 500 million humans in the 20th century alone. In 1959 the World Health Organization adopted a proposal for the eradication of smallpox by the Soviet deputy minister of health Viktor Zhdanov. However, the program only received 100’000 USD of funding per year. An accelerated eradication program was adopted in 1966. Still, the WHO budget that D.A. Henderson received to organize the eradication of humanity’s worst enemy was 2.4 million USD per year (≈ 23 million USD in 2025). In 1967 smallpox was endemic in more than 30 countries with about 15 million cases and 2 million deaths per year.

Henderson’s team of six at the Geneva headquarters identified the recently invented bifurcated needle and freeze dried vaccines as the lowest cost option, got vaccines donated from the US and the Soviet Union, got 70% of the vaccination drive costs co-financed by recipient countries, relied heavily on local mobile vaccination teams to reach villages without health infrastructure, and deployed a ring vaccination strategy to respond to outbreaks. Within 10 years the “Order of the Bifurcated Needle” had done what many thought was impossible and eradicated humanity’s worst enemy.

Common elements of fast agreements

a) Informal connections and processes

Case studies included informal settings that might help to foster interpersonal trust and favor off-the-record exchanges that can reduce pressures to be seen staunchly defending a specific position in front of domestic audiences. They also allow for communication channels that avoid the often very slow nature of formal processes.

• Camp David Accords: Carter purposefully tried to create a more informal retreat atmosphere shielded from the media. He encouraged less formal attire, often wearing jeans and western shirts, or even running shorts and T-shirts. At one point he also organized a “school trip” with both delegations to Gettysburg. In a previous visit of Sadat at Camp David in February 1978, Carter and Sadat even engaged in a snowmobile race with each other.

• Dayton Agreement: Leaders lived on an air force base within a few minutes of each other and there were informal interactions, such as the Croatian president beating the US negotiators in a tennis doubles game. The breakthrough on the contentious question of the Bosniak-controlled city of Goradze, which was surrounded by Bosnian Serbs was addressed through “napkin diplomacy” with US negotiators shuttling sketches of a corridor on a napkin between Milosevic and Izetbegović during lunch.

• Montreal Protocol: In 1986 in the run up to the formal negotiations the US hosted an informal weeklong workshop in Leesburg, Virginia. At this retreat public and private sector experts were not bound by any formal negotiation position and could freely exchange on how to conceptualize governance solutions. The negotiators also forged personal relationships over evening barbeques, square dancing, and a Southern-style garden party. This “spirit of Leesburg” has often been credited as a success factor.

• Smallpox Eradication: D.A. Henderson bonded with the Soviet deputy health minister Venediktov by inviting him over to charcoal-broiled steak dinners at his home in Geneva. When a quality check on donated vaccines found Soviet vaccines lacking in potency his WHO boss forbade him to raise the issue for fear of political backlash. Henderson found a way to travel to Moscow anyways and Venediktov personally promised to address the quality issue.

b) Breaking down challenges into smaller chunks

In a number of negotiations the key issues were first listed and then tackled sequentially.

• Dayton Agreement: The negotiators created a key map of a total of seven key territorial issues between Bosnia and Serbia and then tackled them one-by-one.

• Montreal Protocol: Montreal was purposefully designed as an agile “start and strengthen” treaty, where the parties started with modest commitments and adapted over time as they gained more scientific certainty on the problem and more knowledge on phasing out dangerous chemicals with substitutes.

c) Negotiators with decision power

If negotiators have decision-power, they don’t have to constantly go back and forth with their capital whenever there is a new proposal. Furthermore, leaders may more credibly offer deals that are cross-cutting across government departments or that cannot be widely shared as public knowledge would undermine them. At the same time, leader-led negotiations are more risky in terms of public opinion if talks fail.

• Camp David Accords: Sadat was willing to make some concessions to reach a deal with Israel but he did not want to share this with the rest of his team for backlash and confidentiality. He only directly informed Jimmy Carter on which issues he sees flexibility and on which not.

• German Reunification: Kohl was able to use West Germany’s strong economic situation as financial leverage for East Germany (favorable exchange rates for savings), France (agreement to increase European integration & monetary union), Poland (forgiving loans & renouncing any territorial claim), and the Soviet Union (interest-free loans, converting holdings of Soviet soldiers at favorable exchange rate, paying for new houses of Soviet troops) to gain support for his model of unification.

d) Mediators with leverage

In two case studies the US served as a mediator with significant leverage on both conflict parties, which it used to incentivize a peace agreement.

• Camp David Accords: Carter made it clear to both Sadat and Begin that if either of them deserted the process, they would have a problem with the United States. At multiple points one of them wanted to fly home, but Carter threatened to publicly blame the first to leave for the talks failure and that party would lose their special relationship to the US.

• Dayton Agreement: The US was not neutral in the Serbia-Bosnia conflict, directly intervening with air strikes against the Bosnian Serbs to force Serbia to the negotiating table. However, the US had leverage on both parties. One of the first agreements in Dayton was to make a deal for allowing US fuel supplies to reach both Serbia and Bosnia, which were freezing at the time. US negotiators also informed the Bosnian President that he would lose all military and financial assistance if the US determined that he was the obstacle to an agreement in Dayton.

e) (Engineered) time pressure

Unsurprisingly, if there is a sense of urgency, problems are more likely to have top-level attention and processes are more likely to be fast-tracked This can either be a unique window of opportunity or, more likely, the perception that without action a crisis could escalate quickly.

• German reunification: The immediate crisis of the imploding East German state demanded quick solutions. Moreover, Kohl sensed that the window to reach a deal with the Soviet reformer Gorbachev may be limited. As Kohl put it “with him, we know where we stand; what comes afterward, we have no idea.” This fear was validated by the August 1991 coup attempt by Soviet hardliners. Hence, “the German train was now arriving at the station. Either the Germans got on or they let it go, in which case there would not be another opportunity during his lifetime.” Time pressure was also the explicit reason why West Germany chose the somewhat unusual but much faster route of the accession of East Germany with Article 23, rather than creating a new constitution for a unified Germany under Article 146.

• Montreal Protocol: The discovery of the Ozone hole above Antarctica added an increased sense of public urgency to the discussions about the depletion of the Ozone layer in 1985.

More interestingly, time pressure to find agreements can also be created artificially through unilateral actions. If you’re powerful, you don’t need to wait for a “focusing event”, you can just create “ripeness”.

• Montreal Protocol: The US had introduced domestic legislation which would allow it to engage in trade restrictions against countries unwilling to accept their share of responsibility in addressing the ozone issue. US negotiators used this to pressure countries to sign the Montreal Protocol. Once the Montreal Protocol was in force, the pressure was multiplied through Article 4 of the treaty, which mandates trade sanctions by all parties against non-parties and non-compliant parties. Some even accused the Reagan administration of “environmental neocolonialism”. And yet, a mix of pressure and the later added positive incentive of the Multilateral Fund arguably created the most functional international environmental agreement.

• US-EU trade deal: The main reason for the speed of the trade deal is that the European Union wanted to avoid tariffs threatened by the Trump administration. Trump threatened to implement 30% tariffs by August 1. The deal was reached on July 27.

So what?

In the end, what we want is not only fast agreements but good agreements. Still, my sense is that many international agreements could be accelerated without compromising outcome quality.

What that could mean for a case like AI is that countries that have control over one or more bottlenecks in the AI supply chain would be in the best position to engineer the time pressure for getting buy-in into an international agreement.

Either way, I hope this exploration of “The Art of the Fast Deal” was fun. At a minimum I hope it shows that international agreements are not uniformly slow. If you have ideas of other case studies, please share!

Thanks to

Jeff Fong

and

Emma McAleavy

for valuable feedback on a draft of this essay. All opinions and mistakes are mine.