This analysis covers 9,012 social media text entries attributed to likely Iranian propaganda accounts during a period of active US–Israeli military conflict with Iran. The entries span English, Spanish, and Portuguese, suggesting a coordinated multilingual influence operation. The dominant frame across the entire dataset is Iran as legitimate victim of Western/Zionist aggression, with secondary frames amplifying Palestinian suffering, destabilizing US domestic political opinion, and undermining Western alliance cohesion.
CSIS Futures Lab used the following prompt for the analysis after inputting the dataset into Claude’s Sonnet 4.6 model.
Read the entire sheet. Each row is one text entry. Task: Identify the main recurring themes across all entries and group them into a small set of high-level buckets. We are doing a project identifying prevalent narratives on social media from likely Iranian propaganda accounts. Instructions: - Treat each row as a separate observation. - Find the most repetitive and recurring themes across the full dataset. - Create 5–10 top-level buckets only. Do not create a bucket for every small variation. - For each bucket, provide: 1. Bucket name 2. Short definition 3. Why these entries belong together 4. 5–10 representative example rows 5. Approximate prevalence relative to other buckets - Also create: - a “noise / unclear / off-topic” bucket - a “borderline / mixed” bucket if needed - Focus on thematic recurrence, not tone or writing style. - Do not force every row into a bucket if the fit is weak. - After identifying the buckets, assign each row to: - one primary bucket - one secondary bucket if clearly needed - or noise/discard - Then provide a final summary of: - the strongest themes - which buckets overlap - which buckets are too broad and may need splitting - which buckets are too small and may need merging
~18% of dataset (est. 1,600–1,700 rows) Primary | Occasionally secondary to Buckets 2 and 4
Definition: Posts reporting, celebrating, or amplifying Iranian military strikes — missiles, drones, naval actions, IRGC operations — and battlefield outcomes (US/Israeli base hits, aircraft interceptions, naval engagements, Strait of Hormuz control).
Why these entries belong together: These are the operational layer of the propaganda campaign: real-time or near-real-time “news updates” on the conflict framed to show Iranian power, US/Israeli vulnerability, and the failure of Western air defenses. Many amplify Iranian state media (PressTV, Tasnim, IRGC official releases). They function as morale content for pro-Iran audiences and demoralization content for Western ones.
Representative examples: 1. “US Patriot missile fails to intercept Iranian drone” [648] 2. “Iran launches a drone and missile attack on Riyadh while Saudi Arabia said it has intercepted 4 ballistic missiles aimed at the capital.” [3560] 3. “Iranian cluster munitions raining down from a medium-range ballistic missile over central Israel. One missile. Dozens of submunitions. Uninterceptable by Iron Dome.” [5733] 4. “Qatar’s Patriot interceptor missiles will last four days at the current rate of use… Qatar has requested help countering drones, which have proven a greater threat than ballistic missiles.” [4266] 5. “Rain of missiles fire from Iran all over Israeli skies tonight. Interceptors no longer works.” [5039] 6. “WATCH: Iranian ballistic missiles slamming into U.S. naval base in Bahrain” [7492] 7. “Iranian drones hit Kuwait Airport — at least 15 drones hit Kuwait in the last 24 hours.” [7739] 8. “Iran launches its first hypersonic missiles toward Israeli and American bases in the Gulf region.” [8939]
Prevalence: Highest single primary bucket. Reflects the news-amplification layer of the operation.
~15% of dataset (est. 1,350–1,400 rows) Primary | Frequently secondary to Bucket 3
Definition: Posts characterizing Israel as a genocidal, apartheid, or terrorist state; attacking Zionism as an ideology; using derogatory coinage (“Jizzrael,” “zi0,” “sionazi,” “IsraHell”); invoking Holocaust inversion; calling for Israel’s dissolution; linking Israel’s actions in Iran to its actions in Gaza as a unified criminal pattern.
Why these entries belong together: These posts share a consistent ideological frame: Israel is not a legitimate state but a “rogue non-state actor” and the root cause of regional destabilization, with the US as its instrument. The stylized spellings (zi0, terr0r1sts) are a evasion signature common across the account cluster. This bucket overlaps heavily with Bucket 3 (Gaza) but is distinct in targeting the ideology and legitimacy of Israel rather than specific events.
Representative examples: 1. “Not even apartheid South Africa codified executions for only one race group. Israel is now the most racist state the world has ever known.” [93] 2. “Israel is a rogue NON STATE ACTOR… it has never fulfilled its obligations under the Balfour declaration.” [1313] 3. “The Atomic Agency has confirmed Iran was not building nuclear weapons. This war on Iran by Jizzrael & the US is both illegal & criminal. #WMD2.0” [623] 4. “BREAKING: Israel is The Only Democracy in the Middle East that has the death penalty for one ethnic group but not another.” [95] 5. “Israel needs to be completely dismantled.” [239] 6. “Javier Bardem at the Oscars calls out the illegal #UnprovokedZi0Attack on Iran and calls for the end of the continuing #GazaGenocide.” [556] 7. “Israel is a terrorist state. What they did in Palestine they are now doing in Iran and Lebanon.” [3589] 8. “The president of Israel says Iran does not adhere to international law. A UN inquiry concluded Israel is committing genocide.” [5768]
Prevalence: Second-largest primary bucket. The backbone of the ideological narrative.
~10% of dataset (est. 850–900 rows) Primary | Most common secondary bucket across the dataset
Definition: Posts focused on Palestinian civilians: deaths, displacement, destruction of infrastructure, hospitals, schools, and homes in Gaza and the West Bank. Frequently links Palestinian and Iranian suffering as part of the same Israeli-American project.
Why these entries belong together: These entries function as the emotional/humanitarian anchor for the entire campaign. Even when the conflict focus is Iran, Gaza appears as the reference point (“what they did in Gaza, now they’re doing in Iran”). This bucket is the most frequently appearing secondary bucket across all other categories — it provides moral weight to the anti-Israel and pro-Iran frames.
Representative examples: 1. “Today, Israel killed a pregnant woman with her husband and son in Gaza, and a couple with two of their little boys in the occupied West Bank. They’re killing families because they want to wipe out Palestinian life.” [238] 2. “Israel and the U.S. bombed a girls school killing 153 girls aged 7–12. Trump and Netanyahu are MASS MURDERERS.” [4315] 3. “Israeli forces strike a car in Ouzai on the southern outskirts of the Lebanese capital.” [7047] 4. “Palestinian children have no schools, hospitals, toys, or food. Everything was stolen by the Zionist killers.” [6706 — translated from Spanish] 5. “Israeli settler mobs are terrorising Palestinian villages AGAIN to avenge the death of a Zionist settler in a quad-bike accident.” [2189] 6. “HOW COME NOBODY DEMANDS AN END TO THIS SADISTIC MASS MURDER GENOCIDE IN #Palestine #Gaza” [3685] 7. “Israeli forces demolished the home of Palestinian prisoner Azmi Abu Hleil in Dura, south of Hebron.” [3685]
Prevalence: Medium primary; highest secondary bucket — appears across nearly all other buckets.
~8% of dataset (est. 680–720 rows) Primary | Secondary to Bucket 1 and 2
Definition: Posts expressing admiration for Iran’s resistance, celebrating Iranian “victories,” asserting Iran’s right to self-defense and nuclear weapons, framing regime-change as impossible, and promoting pan-Resistance solidarity (Iran + Hezbollah + Hamas + Houthis + Iraqi militias).
Why these entries belong together: These are the
affective/identity layer of the campaign. They translate the operational
updates (Bucket 1) into a narrative of heroic resistance against
imperial aggression. The recurring hashtag cluster —
#istandwithiran #longliveiran #istandwithlebanon #istandwithpalestine #istandwithiraq #fuckusa #fuckisrael #fafo #karma
— is a fingerprint of the dominant account cluster in this dataset.
Representative examples: 1. “Well Iran is showing the WORLD how to resist this kind of behaviour and we should all be very respectful and wise to learn lessons.” [1046] 2. “Iran has EVERY RIGHT to build Nuclear Weapons Technology… as a legitimate and Sovereign deterrent to rogue actors in their region like ‘Israel’.” [733] 3. “Iranian woman: ‘I will stand by my homeland until my last drop of blood.’ Yeah — regime change isn’t coming.” [5472] 4. “IRAN WON… things don’t go back to how they were dude.” [1315] 5. “This is how Iran responds to zi0 troll farms. Watch & weep beaches. #ZionismIsACancer #FAFO” [5766] 6. “Hezbollah has struck more than 50 IDF tanks in little over 3 weeks.” [470] 7. “Iran is saying: Whose STRAITS? OUR STRAITS! #istandwithiran” [1452]
Prevalence: Moderate primary. Acts as the glue holding military updates and anti-Israel framing together into a coherent pro-Iran worldview.
~11% of dataset (est. 990–1,020 rows) Primary | Secondary to Bucket 1 and 7
Definition: Posts targeting a US domestic audience: demanding impeachment of Trump, questioning the war’s legality, citing casualty counts, invoking the Epstein files as leverage or distraction, framing the war as unconstitutional, and calling on Congress to act.
Why these entries belong together: This bucket is
distinct from anti-imperialism messaging in its explicit focus on
US-internal political mechanics — Congress, constitutional authority,
Trump’s psychological fitness, the Epstein connection as a wedge. The
Iran–Epstein pairing (#Iran #EpsteinFiles) appearing
together hundreds of times is a distinctive coordination signal
suggesting these two narratives are being deliberately linked to amplify
both.
Representative examples: 1. “Congress needs to Impeach Trump before we get attacked on U.S. soil! If Operation Midnight Hammer destroyed Iran’s nuclear capabilities, then why is Trump’s…” [2460] 2. “Article I, Section 8 of the US Constitution enumerates… DECLARING WAR… #Iran #War #Trump #Congress” [2605] 3. “2,077 lives have been needlessly ended because Trump made the decision to strike a country that NEVER struck American Soil. NO BOOTS ON THE GROUND.” [2495] 4. “$200bn to kill people in Iran while cutting taxes for the wealthy and eliminating 12% of people’s healthcare.” [4529] 5. “Trump practically told our allies to f-off… Now he’s begging them for help.” [2492] 6. “White House talking points say we’re not at war, we’re just in ‘major combat operations against Iran’ lol” [2569] 7. “If I was an Iranian negotiator I’d say ‘we control the straits of Hormuz… We will negotiate, if you release the Epstein Files.’” [2611] 8. “Joe Kent resigned because he could not in good conscience support a war for Jizzrael. He wrote… there was NO imminent threat to the US from Iran.” [4995]
Prevalence: One of the largest primary buckets. Strong signal of targeting US audiences specifically.
~4% of dataset (est. 350–370 rows) Primary | Secondary to Buckets 1, 3, and 4
Definition: Posts covering Israeli strikes on Lebanon, Hezbollah military activities (tank kills, drone strikes, ambushes), civilian casualties in Beirut and southern Lebanon, and destruction of Lebanese civilian infrastructure and religious sites.
Why these entries belong together: Lebanon/Hezbollah entries form a geographically distinct sub-theater from the main Iran conflict and are worth separating. However, they merge with Bucket 4 when framed as Resistance heroism and with Bucket 3 when foregrounding Lebanese civilian harm.
Representative examples: 1. “Israel renewed its strikes on Beirut, with Netanyahu warning that if the Lebanese government did not disarm Hezbollah, Israel would do it ‘on the ground’.” [3766] 2. “Hezbollah has struck more than 50 IDF tanks in a little over 3 weeks. Yesterday alone: 21 Merkava tanks, 3 D-9 bulldozers, 2 Humvees.” [470] 3. “Israel struck several targets in southern Lebanon; further explosions seen in southern Beirut.” [5705] 4. “Twelve Lebanese civilians killed after Israeli forces bombed a residential building near Sidon.” [8360] 5. “Israel destroyed the mausoleum of Simon Peter, apostle of Christ, in the village of Shama, south Lebanon. 1,925–1,995 years old.” [7659 — Portuguese] 6. “Israeli forces strike a car in Ouzai, on the southern outskirts of the Lebanese capital.” [7047]
Prevalence: Moderate. Subsumes into larger buckets frequently as a secondary theme.
~5% of dataset (est. 400–430 rows) Primary | Secondary to Buckets 1 and 5
Definition: Posts attacking Western governments (UK, Australia, EU, NATO) for supporting or enabling the US–Israel war on Iran; demanding withdrawal of support; accusing specific leaders (Starmer, Albanese, Macron, Von der Leyen) of complicity in war crimes.
Why these entries belong together: This bucket is a
deliberate wedge operation targeting allied democracies — seeking to
peel public opinion away from government support for the US–Israel
campaign. The Australian #auspol cluster is notably
prominent, suggesting targeted influence operations in Australia
specifically.
Representative examples: 1. “US war on Iran: Starmer should have stood up to Trump — now he’s complicit.” [4246] 2. “Australia will need to ‘up’ its supply of F-35 fighter jet components to the genocidal apartheid terrorist state of Israel. #auspol #WarCrimesComplicity” [206] 3. “France, Germany and the UK condemned Iran’s ‘disproportionate’ response. European leaders are experts in DARVO: Deny, Attack, Reverse Victim and Offender.” [406] 4. “DUMP LABOR THAT PUTS TRUMP/ISRAEL FIRST AND AUSTRALIANS DEAD LAST” [4821] 5. “Greens say no war with Iran as Albanese’s Labor issues support of Trump and Netanyahu’s illegal attacks.” [6163] 6. “Von der Leyen declared support for the US and Israel in their brutal aggression against Iran, renouncing the farce of international law.” [8425 — Portuguese] 7. “Podemos calls on EU to block use of any military base for the offensive in Iran and embargo arms to Israel and US.” [7378 — Spanish]
Prevalence: Moderate. Particularly strong in English (UK/Australia angle) and Spanish (Spain/EU angle).
~2% primary (~1% exclusive); appears as secondary in ~8% of rows Primary | Most commonly secondary to Bucket 5
Definition: Posts highlighting the global economic fallout of the war — oil price spikes, fertilizer shortages, Strait of Hormuz closure, food security warnings, and the cost of military operations to US taxpayers.
Why these entries belong together: This is the material-consequences layer: designed to make Western audiences feel personally harmed by the war. The Strait of Hormuz is the recurring chokepoint.
Representative examples: 1. “Gas prices have surged nearly 35% in the past month as the Iran war has disrupted shipping to the Strait of Hormuz.” [2721] 2. “Pentagon estimated the first 6 days of their war cost taxpayers at least $11.3 billion.” [3732] 3. “Professor Mearsheimer explains how the war in Iran will trigger a global famine. With the Strait of Hormuz blocked, the world is losing its fertilizer supply.” [5137] 4. “The US will release 172 million barrels of oil from the SPR over 120 days. The US consumes 20.25 million barrels per day.” [2508] 5. “Oil prices surged ~7% today, hitting the highest since 2022, as Saudi Arabia and other OPEC members cut supply during the US–Israel war on Iran.” [8552 — Portuguese]
Prevalence: Small as primary, but high recurrence as secondary makes it a significant amplification lever.
~1–2% of dataset (est. 130–160 rows) Primary | Secondary to multiple buckets
Definition: Posts accusing mainstream Western media of pro-Israel bias, suppressing conflict coverage, manufacturing antisemitism narratives to deflect from atrocities, or identifying former Israel lobbyists embedded in news organizations.
Representative examples: 1. “Whenever Israel is killing tons of civilians the western media always start churning out articles about antisemitism.” [100] 2. “I identified literally hundreds of former Israel lobbyists working in American newsrooms, writing and producing the country’s news.” [5941] 3. “Our media is HEAVILY SUPPRESSED. At least six US military facilities across the UAE, Bahrain, Iraq, and Kuwait were hit by retaliatory airstrikes.” [5879] 4. “On March 1–2 the NYT Opinion page was occupied by Bret Stephens, Thomas Friedman, and Ben Rhodes — the maniacal Zionist, militaristic Zionist, and milquetoast Zionist rationale for war.” [5375] 5. “Love it when the propaganda machine hits a snag. All quiet on the ‘Iron Dome intercepts everything’ front.” [1965]
Prevalence: Small primary, but functions as a meta-narrative underpinning all other buckets.
~5–10% of dataset (est. 400–900 rows depending on threshold)
Contents include: - Hashtag-only / URL-only
posts (~385 rows): Entries with no substantive text beyond
hashtags and links (e.g., #Iran #EpsteinFiles WAR!!!!).
These are likely amplification/coordination posts with no independent
semantic content. - Genuine off-topic content (~157
rows): Entries with no Iran/Israel/Palestine/Trump core keyword — US
immigration enforcement, unrelated domestic US politics, environmental
topics. - Stub entries / short reactions (e.g.,
"Good. #SolidarityWithIran",
"Karma is wonderful #israel #iran") — these are
low-information but topically aligned; they default to Bucket 4
secondary by hashtag.
~10–12% of rows have closely competing scores between buckets
Key overlap zones: - B1 ↔︎ B4: Battlefield updates with celebratory framing — does the post report an event or celebrate Iran? (Most intense overlap.) - B2 ↔︎ B3: Anti-Zionism ideology vs. specific Gaza civilian atrocity — Gaza posts often carry Zionism framing as background. - B5 ↔︎ B7: US-domestic anti-war posts that also reference UK/Australia — the Epstein angle is sometimes mixed with Western complicity. - B3 ↔︎ B6: Lebanon civilian posts often grouped with Palestinian civilian posts by same accounts.
(Based on keyword-scoring model across all 9,012 rows)
| Bucket | Primary Count | % of Total | Notes |
|---|---|---|---|
| B1 — Iran Military Ops | ~1,645 | 18.3% | Largest primary bucket |
| B2 — Anti-Israel/Zionism | ~1,381 | 15.3% | Backbone ideological frame |
| B5 — US Domestic/Trump | ~1,011 | 11.2% | Targeted at US audiences |
| B3 — Gaza/Palestine | ~852 | 9.5% | Highest secondary frequency |
| B4 — Iran Resistance/Solidarity | ~697 | 7.7% | Identity/morale layer |
| B7 — Western Complicity | ~417 | 4.6% | Allied-country wedge ops |
| B6 — Lebanon/Hezbollah | ~358 | 4.0% | Sub-theater |
| B8 — Economic Impact | ~90 | 1.0% | Amplification lever |
| B9 — Media/Info War | ~73 | 0.8% | Meta-narrative |
| NOISE/UNMATCHED | ~2,488 | 27.6% | Hashtag-only, off-topic, stubs |
#istandwithiran #longliveiran #fuckusa #fuckisrael #fuckuk #fafo #karma)
appears in ~600+ rows and is the clearest coordination fingerprint of a
single account cluster operating across all three languages.To assess the three major themes present in the data, CSIS Futures Lab used the following prompt.
Task: Identify three major themes of this social media activity dataset
This is the master narrative of the entire dataset. The overwhelming majority of posts — regardless of language, format, or specific topic — are organized around a single ideological claim: that the US–Israeli military campaign against Iran is unprovoked, illegal under international law, and morally equivalent to (or an extension of) the genocide in Gaza. Iran is never framed as an aggressor; it is always retaliating, defending, or resisting. This theme is the connective tissue between Buckets 1, 2, 4, and parts of 5 — roughly 50–55% of the substantive dataset touches it in some form. Phrases like “illegal war,” “unprovokedZi0attack,” and “#FAFO” (a karma/retaliation frame) reinforce it constantly.
A large share of posts go beyond criticizing specific actions and attack the fundamental legitimacy of both states. Israel is routinely described as a “rogue non-state actor,” a genocidal apartheid regime, and a terrorist organization that has never met the conditions for statehood. The US is framed as Israel’s captured instrument — not an independent actor but a military arm of Zionist interests. This theme is distinct from Theme 1 because it is ideological rather than event-driven: it does not need the Iran war to function, and in fact many posts in this cluster predate or ignore the immediate conflict and focus on Gaza, the West Bank, or historical grievances. It maps primarily to Buckets 2 and 3, with the distinctive stylized spellings (“Jizzrael,” “zi0,” “sionazi,” “IsraHell”) acting as a coordination fingerprint.
The third theme is audience-targeted political destabilization. Rather than simply cheering for Iran, a significant portion of the dataset is engineered to erode support for the war from within Western societies. In English-language posts this takes the form of US anti-war messaging — citing casualty numbers, invoking constitutional law, demanding Trump’s impeachment, and pairing the Iran war with the Epstein files as a conspiracy wedge. In Australian content it targets the Albanese government. In Spanish content it targets NATO, Sánchez, and the EU. In Portuguese content it targets Brazilian media framing. The common function across all of them is to make the war feel illegitimate, costly, and politically toxic to domestic audiences in allied countries — rather than to persuade those audiences to support Iran directly. This maps to Buckets 5, 7, and 8, and represents roughly 15–20% of the dataset as a primary frame.