Analysis on the Quick: Using ARC to Assess Different Outcomes of Regime Change in Iran

By Adam Siegel on June 19, 2025

Last weekend I spent time doing some citizen geopolitical analysis to try and better understand what possible outcomes there would be to the war between Israel and Iran after Israel launched a strike against various Iranian military and nuclear enrichment assets.

To begin, I started with a high-level query in ARC: "Will the Israeli attacks on Iran on 6/12/25 ultimately result in regime change and what will direct American military involvement be, if any?"

Within 2-3 minutes ARC had completed its initial research and found 35 links to timely and credibly sourced news sources and background material. Once that was completed, I was able to create drafts of possible outcome scenarios, and a breakdown of drivers and indicators that could influence those scenarios. ARC also generated a 1-page research brief (with citations) designed to bring any reader context on the situation.

Here was the initial output with no editing on my part:

Possible Outcome Scenarios

Scenario 1: Prolonged Israeli-Iranian Military Escalation Without Regime Change or Direct US Involvement
After Israel's June 2025 strikes on Iran, both countries engage in continued tit-for-tat military escalation—airstrikes, missile and drone attacks, cyber operations, and regional proxy actions—causing significant disruption, casualties, and regional instability. However, the Iranian regime remains in power, leveraging nationalist sentiment and repression to maintain control. The US maintains a posture of rhetorical support for Israel and increases force protection for its regional personnel but avoids direct military engagement unless directly attacked. Diplomatic initiatives stall, energy markets remain volatile, and the risk of a broader regional war persists but is contained.

Scenario 2: Escalation to Direct US Military Involvement Following Attacks on American Forces or Interests
Iran, or its proxies, retaliate against perceived US complicity by attacking US military bases or personnel in the Middle East. This triggers a direct US military response against Iranian targets, potentially including strikes on Iran’s missile infrastructure, navy, or cyber capabilities. The conflict expands to involve US, Israeli, Iranian, and proxy forces in a broader regional confrontation. While this scenario raises the risk of regime destabilization in Iran, the regime survives the initial conflict phase, though internal unrest may intensify.

Scenario 3: Regime Change in Iran Triggered by Internal Collapse After Military and Economic Pressure
The cumulative impact of Israel’s decapitation strikes, ongoing economic crisis, and rising public dissent leads to a rapid loss of legitimacy and control by the Islamic Republic. Mass protests, internal fractures within the security apparatus, and elite defections accelerate regime collapse. A transitional authority or council emerges, facing significant instability and potential fragmentation. The US does not intervene militarily but increases diplomatic and economic support for a post-regime transition.

Scenario 4: Diplomatic De-escalation and Return to Negotiations
Following initial escalation and international alarm, intense diplomatic interventions by the UN, EU, Gulf states, and possibly China or Russia succeed in brokering a ceasefire or deconfliction agreement. Both Israel and Iran halt major kinetic operations, Iran resumes participation in nuclear negotiations (possibly under duress), and the US leverages backchannels to stabilize the situation. The Iranian regime remains in power, and the risk of direct US involvement or regime change subsides.

These seemed pretty good but they didn't explicitly mention indirect US military involvement, i.e. providing heavy munitions to reach deeply embedded nuclear sites, or intelligence sharing. So I updated the first and second scenario to include those.

Then I asked ArcBot to "suggest additional scenarios that should be considered, but have a very low likelihood of occurring, so it suggested these:

Scenario 5: Regional War Engulfs Multiple States
The conflict rapidly escalates beyond Israel and Iran, drawing in Lebanon (Hezbollah), Syria, Iraq, and the Gulf States. Missile exchanges and proxy conflicts spiral, leading to widespread destruction across the Middle East. International shipping through the Strait of Hormuz is severely disrupted, triggering a global energy crisis. The US, Russia, and China are forced into tense standoffs as they back different regional actors, raising the risk of great-power confrontation.

Scenario 6: Collapse of Multiple Regimes
The conflict and ensuing turmoil trigger regime collapse not only in Iran, but also in other fragile states in the region (e.g., Iraq, Lebanon). Power vacuums are filled by extremist groups, leading to a wave of civil wars and ungoverned spaces, reminiscent of but far larger than the post-2011 Arab Spring chaos.

Scenario 7: Use of Weapons of Mass Destruction
In a desperate move, either Iran or Israel employs chemical, biological, or even tactical nuclear weapons, either directly or through proxies. This leads to catastrophic casualties, mass refugee flows, and an international emergency response. The taboo against WMD use is shattered, and global security norms are destabilized.

Scenario 8: Internal US Political Crisis Due to Escalation
US involvement in the conflict (whether by choice or after an attack on its forces) leads to severe political polarization and instability at home. Massive anti-war protests, cyberattacks targeting US infrastructure (originating from Iran or proxies), and a surge in domestic terrorism or unrest paralyze the US government’s ability to respond coherently, compounding the global crisis.

Scary stuff, but I would agree with ARCBot's assessment these are pretty low or remote likelihood.

Drivers and indicators that would impact the outcome of the scenarios

After I completed the scenarios, I asked ARCbot to define drivers and indicators I should be watching to better understand looking into the future which scenario is most likely to play out. ARC can leverage the entire knowledge corpus of the OpenAI ChatGPT 4.1 LLM alongside the curated sources ARCbot had found, it produced this outline:

Resilience and Stability of the Iranian Regime
- Speed and effectiveness of leadership replacements after Israeli strikes
- Scale and intensity of domestic protests or unrest in Iran
- Rate of defections or purges within Iran’s security forces and elite circles
- Implementation of martial law or expanded security crackdowns
- Public sentiment in Iran (measured via social media, polling, or emigration rates)
Scope and Nature of Iranian Retaliation to Israeli Strikes
- Number, scale, and success rate of Iranian missile/drone attacks on Israel and other targets
- Involvement of Iranian proxies (Hezbollah, Houthis, Iraqi militias) in regional attacks
- Targeting of US or allied assets/personnel by Iran or its proxies
- Statements from Iranian leadership signaling escalation or restraint
- Evidence of Iranian attempts to close the Strait of Hormuz or disrupt Gulf shipping
Thresholds for Direct US Military Involvement
- Any confirmed attacks against US forces, bases, or diplomatic facilities in the region
- Changes in US military posture (deployments, force protection, evacuation of personnel)
- US government statements shifting from restraint to warnings/threats of retaliation
- Congressional debate or authorization of military force against Iran
- Allied (NATO, Gulf, Israel) calls for direct US military support
International Diplomatic Efforts and Regional Mediation
- Emergency UN Security Council sessions and resulting resolutions
- Mediation efforts by third parties (Oman, Qatar, Russia, China, EU) and their outcomes
- Resumption or breakdown of Iran-US or multilateral nuclear talks
- Adjustments to sanctions regimes or offers of incentives for de-escalation
- Public positions of key regional states (Saudi Arabia, UAE, Turkey) on further escalation or ceasefire
Internal Political Dynamics in the US and Israel
- US presidential and congressional statements on intervention and support for Israel
- Domestic political pressure in the US (protests, opinion polls, partisan divisions) regarding escalation
- Israeli political unity or dissent over continued operations and war aims
- Legislative actions in the US or Israel related to war powers or military funding
- Media narratives framing the conflict and shaping public opinion in both countries

This was all very helpful for an immediate snapshot and frankly much more helpful than reading the various geopolitical think tank and consulting newsletters I subscribe to, but lately we've been thinking about how to take the next step to start tracking an issue like this on an ongoing basis. We happen to have deep expertise in human crowdsourced forecasting, so we've made that capability available, and have also instructed the LLM’s to make forecasts using best practice forecasting techniques.

For example, in my ARC-generated analysis about Israel, based on special instructions on forecast best practices, the LLM assessed the likelihood of my scenarios as follows. Note that I did not instruct ARC to make these scenarios mutually exclusive, so it did not assess the likelihood against one another.

Possible Scenario	Probability Range
Scenario 1: Prolonged Israeli-Iranian Military Escalation Without Regime Change or Direct US Involvement	55% - 80%
Scenario 2: Escalation to Direct US Military Involvement Following Attacks on American Forces or Interests	55% - 80%
Scenario 3: Regime Change in Iran Triggered by Internal Collapse After Military and Economic Pressure	20% - 45%
Scenario 4: Diplomatic De-escalation and Return to Negotiations	20% - 45%
Scenario 5: Regional War Engulfs Multiple States	5% - 20%
Scenario 6: Collapse of Multiple Regimes	5% - 20%
Scenario 7: Use of Weapons of Mass Destruction	5% - 20%
Scenario 8: Internal US Political Crisis Due to Escalation	0% - 5%

ARCbot also generated forecast questions I can use to solicit human input (and soon, LLM input!) I can publish these questions to whatever site or distribution list I want, or to an existing Cultivate forecast platform. Here are a few that were automatically generated for me that are live for forecasting (go ahead, try one!)

To use these I would likely change some timelines and streamline the wording, but they were a good first draft!

Of course I can also leverage external sources, like RAND’s Forecasting Initiative. Here’s the current forecast for a question they’re asking about America’s involvement:

Will the United States launch a kinetic strike against Iran before 1 August 2025?

Source: RAND Forecasting Initiative (RFI)

With these kinds of inputs I can continue to track which scenario is most likely to play out or iterate on the scenarios themselves by asking ArcBot to update them based on the latest source content it has gathered.

Our vision for ARC is to use AI agents and open source and proprietary datasets to monitor, forecast, and update all indicators and scenarios while still facilitating human iteration and collaboration for the analysis, then giving policy and decision makers a robust toolset of decision support that would leverage this analysis.

Ultimately using ARC will mean decisions and policies can be made under less uncertainty and much more quickly based on credible analytic techniques. And from a business perspective, if in 5-10 minutes someone can get to analysis that is largely on target and insightful, then start exploring decision options with the awareness of an organization's strategy and existing capabilities, the amount of coverage a single person can have can be dramatically scaled and be at a higher quality than it would have otherwise.

In practice this means much more time could be spent doing deeper, more nuanced research, focusing on collecting information where the AI can’t, or working through decision options — all very high-value activities that will fundamentally improve an organization's or government’s competitive, market, or security stance, all while realizing a significant ROI.