Method & integrity. Every claim is tiered — confirmed (primary document / multiple reputable outlets), reported (single credible outlet), alleged (claimed, contested, or unproven) — and every tactic is classified into one of four buckets, because conflating them is how this kind of work gets discredited:
- DISINFORMATION — demonstrably false content, deliberately spread.
- MANIPULATIVE-BUT-LEGAL — deceptive design/targeting that isn't false (dark posts, dark patterns, fake-persona outreach, AI-seeding).
- AGGRESSIVE-LEGITIMATE — hardball but truthful (industrial A/B testing, microtargeting on lawful data).
- MYTHOLOGY — widely claimed, NOT supported by the record.
Built from public sources only; allegations are labelled, not asserted. Not legal advice.
Executive synthesis — the honest headline
Across a 20-year career, Parscale's documented record is overwhelmingly manipulative-but-legal political technology, not a trail of fabricated "disinformation." He is a builder of targeting and persuasion machinery — sold to whoever pays (the GOP, Argentina's Milei, Bolivia's Paz, Honduras's Asfura, the State of Israel). The machinery has grown steadily more deceptive at the edges — from 2016 dark posts, to dark-pattern-adjacent fundraising, to 2020 doctored attack ads, to 2025–26 fake-persona texting and AI-output seeding.
But the strict-sense disinformation in his orbit largely belongs to others: the Russian IRA (separate, no coordination — Mueller), the WinRed dark patterns (built by Coby/Lansing, not him), the AMMC "$617M shell" (Kushner-controlled), and the Brazil "stolen election" hoax (his business partner Fernando Cerimedo's act, not his). The genuinely documented false/manipulated content attributable to his own operation is narrower and specific (2020 doctored Biden ads; census-impersonation ads Facebook pulled), and the genuinely novel and current item is the FARA-registered Israeli AI-influence apparatus — disclosed, self-dealing, built to bend AI, of unproven effect.
The strongest, most defensible characterization is therefore: a foreign-and-domestic political-influence contractor whose persuasion machine is increasingly deceptive by design, who now works for a foreign government, and who keeps company with documented disinformation operators — but whose own provable "disinformation" acts are fewer than the legend claims. That is harder to dismiss precisely because it doesn't overreach.
1. 2016 — Project Alamo
- confirmedAGGRESSIVE-LEGITIMATE San Antonio digital hub (Giles-Parscale, ~100 staff), ~$94M through the firm; Facebook ~80% of digital budget; raised $250M+ small-dollar. Industrial A/B testing: 50,000–60,000 ad variants on an average day, ~175,000 on the third-debate day (RNC's Gary Coby). (CBS 60 Minutes; Wikipedia/Project Alamo.)
- confirmedMANIPULATIVE-BUT-LEGAL "Dark posts" — targeted ads visible only to the recipient, hiding messaging from public scrutiny. Legal then; the opacity is the concern (Facebook later required public ad archives).
- [confirmed (existence) / reported (intent)]MANIPULATIVE-BUT-LEGAL Voter "deterrence": a senior official told Bloomberg "we have three major voter suppression operations under way" (idealistic white liberals, young women, African Americans). Channel 4's leaked dataset (2020) found ~3.5M Black Americans tagged "Deterrence," over-represented vs. population. But the content used was true-but-weaponized (the "superpredator" ad used Clinton's real 1996 words) — manipulative targeting, not fabricated disinformation. The causal claim that it lowered Black turnout is alleged, unproven (multi-causal).
- confirmed No demonstrably-false fabricated content is attributable specifically to Parscale's 2016 operation.
2. Cambridge Analytica — de-mythologized
- confirmed CA was a paid vendor: $5,912,500 total (FEC: the $5M payment was 1 Sept 2016), ~a dozen staff embedded under Matt Oczkowski.
- confirmedMYTHOLOGY "Psychographics won 2016." Overstated and effectively debunked — Parscale, Oczkowski, and Nix himself later said psychographics weren't deployed in the general; the campaign switched to RNC data; Kogan called the method "a dead end"; the UK ICO found CA "dramatically overstated" its capabilities.
- confirmedMYTHOLOGY "Trump targeting ran on the 87M harvested Facebook profiles." The harvest was real (Kogan/GSR) — but it was CA's data wrongdoing, and the ICO, holding the seized servers, could not determine the data was used in any specific campaign. Parscale's ads used Facebook's standard lawful Custom/Lookalike Audiences on RNC data. Do not fuse the scandal (illicit harvest) with the campaign (lawful ad tools).
3. 2020 — the campaign he actually ran (Feb 2018–July 2020)
This is where the documented false/manipulated content tied to his operation lives:
- confirmedDISINFORMATION "Official 2020 Congressional District Census" ads impersonating the US Census — Facebook removed them.
- confirmedDISINFORMATION 1,100+ ads with a digitally-altered photo of Biden with an added earpiece (4.7M+ impressions); "satire" defense rejected by fact-checkers.
- confirmedDISINFORMATION A deceptively edited Biden video ("circular firing squad" cut to fake a Trump endorsement) — Twitter's first "manipulated media" label.
- confirmedDISINFORMATION Edited "Biden in a basement" photo (altered from an outdoor Iowa event); the misleading "China-soft Biden" ad (Gary Locke image).
- confirmedMANIPULATIVE-BUT-LEGAL The debunked Ukraine/"Biden corruption" ad that Facebook refused to remove under its political-ad fact-check exemption — the central case in that policy fight.
- confirmed Tulsa (June 2020): "over 1M ticket requests / biggest data haul of all time" vs. ~6,200 actual attendance; TikTok/K-pop mass-registration; he dismissed the prank.
- confirmed Demoted 15 July 2020 (Bill Stepien in); drivers: Biden's polling lead, Tulsa, resentment over ~$39M to his firms and the ~$1B burn.
- Attribution caveat: these ads ran on his infrastructure during his tenure; individual creative decisions are rarely traced to him by name.
4. WinRed dark patterns — NOT his (the key attribution)
- confirmed[MANIPULATIVE-BUT-LEGAL — donor deception, not disinformation] Pre-checked recurring + "money bomb" boxes drove ~$122.7M in refunds (10.7% rate vs. ActBlue's 2.2%); NYT's Stacy Blatt investigation (Apr 2021).
- confirmed Built/operated by WinRed (Gerrit Lansing) + campaign digital director Gary Coby — NYT called it "a Gary and Gerrit production." Parscale is not named as the author, and the scheme peaked Sept–Nov 2020, after his July demotion.
- confirmed Parscale's documented tie: he was "involved in creating" the WinRed platform (2019) and pushed its adoption — he did not author the deceptive boxes and does not own WinRed. Campaign Nucleus didn't publicly exist until ~Dec 2020 — it had no role in the UX.
- confirmed Adjudication: 4 state AGs investigated; WinRed's preemption suit lost (8th Cir., Feb 2023 — states may investigate); FEC only recommended a ban; no federal ban, no FTC action. The "$135M" figure sometimes cited is unverified; the real numbers are $122.7M total / $64.3M final stretch.
5. Campaign Nucleus & the ad-tech ventures — the AI reality check
- confirmed Campaign Nucleus = real conservative campaign SaaS (CRM, email, fundraising, "war room"). Clients: Trump 2024 ("Powered by Nucleus"), reportedly Mike Johnson; >$2.2M from Trump/RNC/PACs.
- [alleged — self-reported][MYTHOLOGY-adjacent] Every efficacy/scale figure is the company's own (14B → now 51.6B+ emails; "$119M raised"; "persuades ChatGPT"). Independent reality check: Trump 2024 publicly said it did not use any AI company's tools; Axios could not reproduce Parscale's claim that ChatGPT cites his sites. The "AI" layer is consistently the least-demonstrated, most-marketing part of every venture.
- reported Marketed capability to build pages "designed to look like legitimate media outlets" — a documented concern, not a confirmed deployed fake outlet.
- [confirmed corrections] "EyesOver" is a pre-existing Canadian firm brought into his orbit, not built by him. No product named "Aiden" exists (likely a misnomer for AiAdvertising's PersonaAI/SWARM/CPP). AiAdvertising (ex-CloudCommerce) collapsed to ~$0.0001/share (June 2026); Facebook removed CloudCommerce ads (2020) over misleading dividend claims.
- reported 2026 "Freedom 250" World Cup Fan Zone on the National Mall funnels registrant name/email/phone through Campaign Nucleus (Daily Caller test-registration confirmed "Powered by Nucleus"). Onward routing to voter files is alleged, inferred, not documented.
6. Foreign elections — Numen & the Cerimedo association
- reported Parscale is a partner in Numen (Buenos Aires) with Fernando Cerimedo, selling data/targeting (Campaign Nucleus + EyesOver) into right-wing Latin American campaigns: Milei (Argentina 2023), Paz (Bolivia 2025 — Parscale traveled there), Asfura (Honduras 2025 — advised remotely).
- confirmed[DISINFORMATION — Cerimedo's, NOT Parscale's] Cerimedo ran the #BrasilFoiRoubado "stolen election" hoax (Nov 2022) via La Derecha Diario; Brazil's electoral court ordered takedowns; he was investigated re the Jan 8 2023 attacks (prosecutor declined charges). This predates the reported Parscale partnership and is squarely Cerimedo's act.
- [confirmed precision] The defensible statement: Parscale is in business with a documented disinformation/election-denial operator and built the targeting infrastructure ("Brad set up all the infrastructure I work with" — Cerimedo). No source places Parscale as producer/director of the false fraud content. Imputing Cerimedo's disinformation to Parscale is unsupported. Bolivia/Honduras "fraud narrative template" claims are alleged, analytic, not artifact-level proof against Parscale.
7. Israel — Clock Tower X / FARA #7649 (the current, novel apparatus)
Fully documented in this folder's DOSSIER.md. Summary:
- confirmed Clock Tower X LLC (Parscale, sole owner) registered FARA #7649 (18 Sep 2025), foreign principal State of Israel via Havas Media Germany; contract $6M→~$9M, ~$15M reported paid, ~$13M disbursed onward (SparkFire ~$6M, Portman Road/Mike Shields ~$5M, Salem Media Reps >$500K, Three Tech ~$500K).
- [confirmed by our own infra check] The 11-site advocacy network is one operator (9 sites on a single Cloudflare account, ~10-min registration bursts 18 Sep / 10 Oct 2025, factsignal.org hub) — and every site carries the FARA disclosure footer (disclosed, not hidden).
- [reported, from the filing's SOW]MANIPULATIVE-BUT-LEGAL Built to influence ChatGPT/Gemini/Claude/Grok via MarketBrew SEO + content ring + Salem authority + SparkFire's chatbot funnel; ≥100 assets + 5,000 variants + 50M impressions/month; ~80% Gen-Z.
- reportedMANIPULATIVE-BUT-LEGAL SparkFire texting uses fabricated personas and non-existent "peace" org names ("Friends for Peace," "Partners in Peace") with no disclosure on the texts — the strongest live FARA-§614(b) adequacy concern (disclosure present on site footers, reportedly absent on texts/social/AI channels). No evidence of impersonating real named individuals.
- confirmed[effect unproven] No model shown to cite the network; no OpenAI/Anthropic deal (Snopes debunked "Israel signed a deal with ChatGPT"). State as intent, not effect.
- confirmed Parscale = Salem Media CSO (Jan 2025). reported Self-dealing: he routed >$500K of Israeli money to Salem Media Representatives — his own employer.
- reported He co-founded Jackson Parker with oil billionaire Tim Dunn (early 2025, Israel-evangelical advocacy); its FARA status is the cleanest open gap (job posting cites "FARA compliance," but no registration found — verify before asserting "unregistered").
- [confirmed — kept distinct] The 303-megachurch geofencing belongs to Show Faith by Works (FARA #7653, Chad Schnitger) — a separate registrant, not Parscale — and was scrapped (Nov 2025) after backlash. Do not attribute it to him.
- reported Tim Dunn ties it together (Hexagon Foundation → ~$70M to WaterStone, which is taking Salem private at a ~250% premium; Dunn also funds Jackson Parker).
9. Adjudications & scrutiny — the outcomes
- FEC MUR 7784 (2020, AMMC + Parscale Strategy, ~$170M alleged mis-disclosed): dismissed (3–3 deadlock → closed), D.C. Circuit affirmed Jan 2024. No violation. AMMC was Kushner/campaign-controlled (Lara Trump, John Pence, Sean Dollman) — not Parscale's firm. ($617M = total throughput; ~$170M = alleged mis-disclosed — keep distinct.)
- WinRed: state-AG investigation upheld (8th Cir. 2023); no FTC action (claims no jurisdiction over campaign fundraising).
- ProPublica "Myths of the Genius" (2019): his skill was management/assembly, not invention; he took commissions on ~$18M in RNC ad billings despite the "no commissions" claim.
- Jan 6: subpoenaed for phone records, not as a witness; released texts show he privately blamed Trump ("guilty for helping him win") and publicly told rioters to stop. No charges, ever, in any matter.
- Sept 27, 2020 incident: Baker Act mental-health hold, 10 firearms secured, no charges; the domestic-altercation report was later walked back by his wife. (Recorded as documented public record only.)
- FARA / Clock Tower X: registered, no DOJ enforcement action to date.
10. Mythology control — the "do NOT overstate" spine
- "He ran Stop the Steal / Jan 6." → False. Demoted July 2020, withdrew Sept 2020 (before the election); privately blamed Trump; told rioters to stop. No operational role in any fraud scheme. (Rhetorically he did echo fraud claims post-election — the one nuance.)
- "CA psychographics won it / he ran on stolen FB data." → Overstated/unproven (see §2).
- "He built the $617M AMMC shell." → Kushner's, not his (see §9).
- "He refused commissions." → False — took commissions on ~$18M RNC billings.
- "Israel signed a deal with ChatGPT via Parscale." → False (Snopes).
- "His AI actually works." → No independent evidence; his marquee client (Trump 2024) disavowed AI tools; Axios couldn't reproduce.
- "Brazil election-fraud disinfo is his." → Cerimedo's act; Parscale built the targeting infra, not the false content.
- "i4i / Aiden / CUI Global are his." → Fabricated associations — unrelated entities.
- Net-worth figures → unsourced aggregators; do not cite.
- Russian IRA "fake news" → separate operation; Mueller found no coordination.
The verdict, in one paragraph
Brad Parscale is the operator of an increasingly deceptive political-persuasion machine — industrial microtargeting and dark posts (2016), a digital operation that ran demonstrably doctored attack ads (2020), a campaign-tech/AI business whose capabilities are real but whose AI claims are unproven marketing, a foreign-election targeting partnership with a documented disinformation operator, and now a FARA-registered foreign-influence apparatus for the State of Israel built to bend generative AI. The throughline is manipulation-by-design and selling that machine to power, foreign and domestic — documented, escalating, and worth public scrutiny. The legend that makes him a master of disinformation and the hidden hand of Stop the Steal is not supported; the truth — a disclosed, for-hire influence contractor whose methods corrode the information commons and who now answers to a foreign government — is more defensible, and damning enough.
Sources: nine research streams with primary FARA filings (efile.fara.gov #7649/#7652/#7653/#7662), FEC records (MUR 7784, FEC API), court records (8th Cir. WinRed, D.C. Cir. CLC v. FEC), and reporting from The Intercept, Responsible Statecraft, ProPublica, NYT, Channel 4, Bloomberg, CBS 60 Minutes, PBS Frontline, FactCheck.org, PolitiFact, Snopes, Axios, DFRLab, El País Bolivia, and others. Full per-claim URLs are in the nine source reports underlying this synthesis.