UK FACT CHECK POLITICS
UK FACT CHECK POLITICS
Voting intention Labour Lab 27% Con Con 21% Reform Ref 20% Lib Dem LD 12% Green Grn 10% Poll tracker
Menu
Account

The Citizen Lab (@citizenlab) on X

x.com 11 November 2026 at 09:10 View original article →

The original link is the material being checked. It may contain false or misleading claims — we link for transparency, not endorsement.

84/100
Verdict High Trust Credible – an accurate summary of a primary research report, with the attribution claim correctly worded as a hypothesis

This is a thread from the Citizen Lab (University of Toronto) dated 3 October 2025.

✓ 5 checks out ✕ 0 disputed ? 3 unverified
☰ Full analysis The complete summary ⌄

This is a thread from the Citizen Lab (University of Toronto) dated 3 October 2025. It promotes the lab's Report No. 189, 'We Say You Want a Revolution: PRISONBREAK', dated 2 October 2025. The descriptive claims in the thread are all backed by the report: more than 50 inauthentic X profiles, AI-generated deepfakes, impersonation of news outlets, and activity timed to the June 2025 Twelve-Day War. Several independent outlets reported the findings (CyberScoop, Ynetnews, Dawn, IBTimes UK, Schneier on Security, AI Incident Database, MIT AI Risk Repository). Independent fact-checkers (France 24, DW via Yahoo, VRT, Hany Farid at UC Berkeley) separately judged the best-known piece of content, the Evin Prison explosion video, to be AI-generated or fake. The weakest point is attribution. The thread says the evidence 'suggests a strong link' to the Israeli state and calls Israel's involvement a 'hypothesis'. That matches the report's own conclusion: the explanation most consistent with the evidence is an unidentified Israeli government agency, or a subcontractor under its close supervision. No independent confirmation was found, and nothing in the research shows the Israeli government commenting either way. The attribution therefore stays Unverified as fact, though it is accurately reported as Citizen Lab's assessment. The only reply in the thread ('Cool story, bro.') is dismissive and makes no factual claim.

What checked out (5)

  • C1: The Citizen Lab published a new report on a coordinated network of fake X profiles spreading AI-generated content to push Iranians towards revolt, and named it 'PRISONBREAK'. Verified by the primary report (Report No. 189, dated 2 October 2025; 'more than 50 inauthentic X profiles'). Sources: https://citizenlab.ca/research/2025-10-ai-enabled-io-aimed-at-overthrowing-iranian-regime/ and https://citizenlab.ca/wp-content/uploads/2026/08/189_We_Say_You_Want_A_Revolution.pdf. Corroborated by https://cyberscoop.com/citizen-lab-disinformation-campaign-israel-iran-evin-prison/ and https://www.ynetnews.com/tech-and-digital/article/rjj7116qpeg
  • C2: PRISONBREAK made and spread deepfake videos and images, plus fake content posing as trusted news sources. Verified. The report says AI was 'routinely used' and records accounts tagging or imitating BBC Persian, VOA Farsi and Iran International. Separately, the Evin Prison gate-explosion video was judged AI-generated or fake by France 24 (https://www.france24.com/en/tv-shows/truth-or-fake/20250630-yes-viral-video-of-tehran-s-evin-prison-explosion-is-ai-generated), DW via Yahoo News UK (https://uk.news.yahoo.com/fact-check-viral-evin-prison-130300345.html), VRT (https://www.vrt.be/vrtnws/en/2025/06/26/broadcast-worldwide-but-fake-video-of-explosion-at-iran-prison/) and UC Berkeley's Hany Farid (https://www.ischool.berkeley.edu/news/2025/hany-farid-takes-look-viral-evin-prison-blast-video-says-its-likely-ai-fake)
  • C3: The network's posting activity lined up closely with the IDF's military campaign against Iranian targets in June 2025. Verified as a finding of the primary report. For example, the first confirmed PRISONBREAK post about the Evin bombing went out at 11:52 a.m. Tehran time on 23 June 2025. The wider context is independently confirmed: the Twelve-Day War began on 13 June 2025 (Britannica: https://www.britannica.com/event/12-Day-War; Al Jazeera: https://www.aljazeera.com/news/2025/6/26/visualising-12-days-of-the-israel-iran-conflict), and Israel struck Evin Prison on 23 June 2025 (Human Rights Watch: https://www.hrw.org/news/2025/08/14/iran-israeli-attack-on-evin-prison-an-apparent-war-crime). Caveat: no outside team has reproduced the timing analysis.
  • C4: The 'Twelve-Day War' (#12DayWar) took place in June 2025. Verified. Britannica gives 13–24 June 2025; Simple English Wikipedia gives 13–25 June 2025. The small discrepancy reflects when the ceasefire is counted as taking effect (24 June).
  • C5 (as a description of the report): Citizen Lab put forward the hypothesis that Israel is behind the network. Verified. The report says 'the hypothesis most consistent with the available evidence is that an unidentified agency of the Israeli government, or a sub-contractor working under its close supervision, is directly conducting the operation.' The thread describes this accurately.
? Unverified claims 3 claims ⌄
  • C5 (as a factual matter): that the Israeli state or its contractor actually runs PRISONBREAK. Unverified. This rests on Citizen Lab's analysis of circumstantial evidence (timing, content and narrative alignment). No admission or denial by the Israeli government was found, and no platform takedown statement from X naming the operator was found. Secondary descriptions also disagree on how confident the lab is: the AI Incident Database says 'medium confidence' (https://incidentdatabase.ai/cite/1221/), while a personal blog says 'high confidence' (https://erkansaka.net/2026/03/23/ai-native-war-disinformation-iran-israel-us/). It should be treated as a reasoned hypothesis, not established fact. A separate investigation by Haaretz and TheMarker into an Israeli-funded Persian-language campaign promoting Reza Pahlavi, reported by Dawn (https://www.dawn.com/news/1946459), makes the hypothesis more plausible but does not prove it for PRISONBREAK.
  • C6: The intent was to cause 'unrest in Iran'. This is an interpretive judgement about motive. It is consistent with the content documented in the report (narratives calling for regime change), but intent cannot be independently checked.
  • C7: That the Evin Prison deepfake came from PRISONBREAK specifically. Independent fact-checkers confirm the video is fake. The claim that this network first posted or amplified it relies only on Citizen Lab's account-level analysis, which no one has independently reproduced.
⌕ Sources & how we checked Search journal, source grades, confidence ⌄

Confidence

High (descriptive claims) / Medium (attribution); overall: Medium-High — The post matches a dated primary research report from a reputable academic lab, co-written with an independent university researcher. Its core descriptive claims (the network, AI deepfakes, news impersonation, timing with the June 2025 war) are supported by that report, by consistent reporting from several outlets, and by independent fact-checks of the key Evin Prison deepfake made before the report. The score is held at 84, not higher, for four reasons. The Israeli attribution comes from a single analytical source with no independent confirmation and no official response found. Secondary sources disagree on the stated confidence level. The thread leaves out the report's caveat that engagement was limited. And the full report was reviewed only through extracts.

Search journal

Citizen Lab PRISONBREAK Iran AI network report

citizenlab.ca 2025/10 AI-enabled influence operation Iran PRISONBREAK

PRISONBREAK fake X accounts Israel Iran deepfake Evin prison

12-day war Israel Iran June 2025

Haaretz PRISONBREAK Israeli disinformation network Iran Citizen Lab

Citizen Lab PRISONBREAK attribution Israeli government subcontractor most consistent hypothesis

Evin prison deepfake video gate explosion BBC Verify AI-generated

PRISONBREAK Clemson Media Forensics Hub Linvill Warren Iran

Israel response Citizen Lab PRISONBREAK report denial comment

Haaretz TheMarker Israeli-funded campaign fake Persian personas Reza Pahlavi

Israeli strike Evin prison 23 June 2025 deaths Iran judiciary

Article metrics

Emotion 7% · Reading grade 14.5 · 166 words

≡ The article we checked Full text as retrieved ⌄
## Post ## Post ## The Citizen Lab on X: "NEW REPORT: We uncovered a coordinated network of fake X profiles that has been spreading AI-generated content to induce revolt in #Iran. We call this network “PRISONBREAK”. Read the report ➡️ ## Conversation * []( NEW REPORT: We uncovered a coordinated network of fake X profiles that has been spreading AI-generated content to induce revolt in [#Iran]( We call this network “PRISONBREAK”. Read the report ➡️ * []( With the help of [#AI]( PRISONBREAK produced and spread deepfake videos, images, and falsified content from trusted news sources - all with the apparent intent of causing unrest in Iran. []( Our analysis of the network’s posting activity shows close alignment with the Israel Defense Force’s military campaign against Iranian targets in June 2025 during the [#12DayWar]( View the timeline: [citizenlab.ca/2025/10/ai-ena…]( " []( Evidence suggests a strong link between PRISONBREAK and the Israeli state, leading to the hypothesis that Israel is behind the network as part of a larger goal of destabilizing the Islamic Republic of Iran. []( * []( Cool story, bro. [Video 3](
You read this free. Help keep it free for everyone. No paywall, no owner with an agenda. Reader support pays for the research.

Get email updates

Our latest reporting and fact checks in your inbox. Free, and one click to leave.

Share this fact check