The Anchor
The anchor at Al-Fassel News doesn’t have a name. She has a face — AI-generated, professional, calibrated to the audience she’s reading to. She reads Arabic-language coverage of Iranian military operations, Houthi strikes, Hamas casualties.
Al-Fassel has an About page. It states that the outlet is “publicly funded from the budget of the United States Government.”
It does not say Central Command. It does not say that the site is one of seven coordinated outlets operating under URLs that all share a “/gc_#/” pattern. It does not say there is no newsroom behind the anchor — no editorial meeting, no beat reporter who built regional sources over years. The AI anchor reads. The U.S. military curates what she reads.
This is what semi-overt propaganda looks like in 2026: real facts, strategic omissions, and a disclosure technically present and practically designed to be missed.
The Network
Between 2022 and 2026, U.S. Central Command built what analysts call the third generation of the Trans-Regional Web Initiative (TRWI) — a military psychological operations program running in various forms since 2008.
The current generation comprises seven confirmed sites, identifiable by the “/gc_#/” structure in their URLs:
- Al-Fassel News (gc_1): Arabic-language, focused on Iran, the Houthis, and Hamas
- Pishtaz News (gc_3): Farsi-language, covering Iran’s military, economy, and internal repression
- Entorno Diario (gc_4): Spanish, English, and Portuguese, targeting Latin America with China-focused reporting
- Kontur Novosti (gc_6): Russian-language anti-Kremlin content for Ukrainian audiences
- GlobalWatch (gc_7): Multi-language security coverage
- BlueShift News (gc_8): English-language space security reporting
- Focus News (gc_9): Indo-Pacific regional security
Each site covers a different geographic and linguistic target. Each carries the same generic funding disclosure. None of their social media accounts consistently carry the state-backed media labels that platform policies nominally require.
In April and May 2026, as U.S. and Israeli forces were conducting operations against Iran, Pishtaz had active Meta advertising campaigns running content about Iranian internet shutdowns and economic distress. GlobalWatch was running ads featuring Ukrainian military imagery. The content was paid media. The source was undisclosed on every channel where the audience actually saw it.
From Bots to Budgets
The shift is worth understanding because it changes the manipulation mechanics entirely.
The first generation of Pentagon influence operations — dismantled by Meta and Twitter between 2019 and 2022 — relied on coordinated inauthentic behavior: fake accounts, bot networks, manufactured engagement. Meta removed those networks. The removal was operationally straightforward: the accounts were fake, and inauthenticity violates terms of service.
The current generation doesn’t violate terms of service. It purchases reach through legitimate advertising infrastructure. Meta’s ad platform does not distinguish between a news organization buying audience and a military psyop buying audience. When you pay for reach, the people who engage are real. The engagement is not manufactured. The comments, shares, and reactions come from genuine users who genuinely responded to the content.
“Paid advertising is not fake or inauthentic engagement,” the Lawfare analysis noted. The audience reactions are authentic. Only the source isn’t.
This distinction matters operationally. Platform trust-and-safety teams are structured to detect inauthenticity. Bot networks leave fingerprints: coordinated posting times, shared IP addresses, activation patterns. Paid advertising leaves a different kind of fingerprint — ad spend, targeting parameters, A/B test history — that is not visible to the audience and is not publicly reviewable.
The new model is harder to attribute, harder to remove, and produces genuine engagement metrics that make the sites appear to have real readership. Because in a functional sense, they do.
What the About Page Says
All seven sites carry a disclosure. The language varies slightly; the structure is consistent: something to the effect of “publicly funded from the budget of the United States Government.”
That disclosure has a specific shape. It confirms government funding. It omits the agency. It omits the program. It omits the purpose.
A reader who finds the About page — itself a low-probability event, since most users encounter these outlets through social feeds — learns that the outlet is U.S.-government funded. They might reasonably assume this resembles Voice of America: a government-funded outlet with an explicit public diplomacy mandate and transparent labeling.
They are not reading Voice of America.
Voice of America carries its identity in its branding, its editorial disclosures, and its stated mission. Its public diplomacy function is explicit. Its audience knows what it is.
The TRWI sites are engineered to occupy the gap between transparent state media and independent journalism. Technically disclosed, they borrow the credibility signals of independent reporting — professional formatting, bylines, sourced quotes, real facts — while suppressing the one piece of information that would change how a reader evaluates everything else: this content was produced to serve a U.S. military psychological operations objective.
Disclosure that does not inform is not disclosure. It is deniability architecture.
The Content Model
The sites don’t fabricate. This is the operationally important point.
Individual claims in Al-Fassel or Pishtaz articles may be true. Iranian internet shutdowns are documentable. Houthi military activities are real. Economic distress in Iran is measurable. The Lawfare analysis of these outlets characterizes their method as “story selection, omission, and attenuated provenance” rather than fabrication.
This model is harder to challenge than disinformation. You can fact-check a false claim. You cannot fact-check a selection — the decision to cover story A and not story B, to include source X and not source Y, to frame a real event in a context designed to produce a particular interpretation.
A reader who encounters an Al-Fassel article and verifies its claims will find them accurate. They will not find the stories omitted because they complicated the preferred narrative. They will not find the attribution needed to properly calibrate the outlet’s perspective. The facts check out. The overall picture is engineered.
Saudi-funded sources appear without context. Anonymously sourced claims run without attribution. Gaza ceasefire dynamics are presented through a specific frame. Iranian military capabilities are described to emphasize weakness. Nothing is false. Everything serves a purpose.
The Influence Tactics Breakdown
Provenance Laundering Through Generic Attribution. “Publicly funded from the budget of the United States Government” is technically accurate and strategically ambiguous. It prevents the operation from making false claims about funding while stripping the specific attribution — CENTCOM, military psychological operations, Trans-Regional Web Initiative — that would allow a reader to correctly identify what they’re reading. The disclosure names the funder without naming the function. A reader who knows Voice of America is funded by the U.S. government will map this disclosure onto that template. The template is wrong.
Legitimate Infrastructure as Credibility Signal. Professional news formatting, AI newsreaders with faces appropriate to target-audience expectations, journalism conventions (bylines, quotes, sourcing), active social media presence — these signals create credibility that functions independently of the actual source. A Farsi-language reader who encounters Pishtaz News has no visual or structural reason to conclude it differs from an independent outlet covering Iranian affairs. Nothing in the presentation signals “military psyop.” The credibility is borrowed from the form.
Paid Amplification as Non-Removable Engagement. Bot networks were removable because they were inauthentic. Paid advertising produces authentic engagement — real users who clicked, commented, and shared. That engagement cannot be reversed. Once a Pishtaz article has reached 40,000 people via Meta advertising and some portion have shared it to their own networks, the distribution continues without the ad spend. The ad is temporary; the reach is durable. A platform can remove a fake account. It cannot un-share an article.
Platform Compliance Gap. Meta and Google policies require state-backed media to carry labels on their accounts. These sites have not consistently received those labels. The gap between the policy and its application is operational runway: the site technically discloses on its website, the platform technically has a labeling policy, and neither mechanism reaches the user who encounters a post in their feed without clicking through to the outlet’s About page. Three layers of technical compliance produce zero effective disclosure at the point of contact.
Omission as the Primary Manipulation Vector. The sites publish real facts. The manipulation is in what isn’t published — the context, the competing narratives, the full attribution to a source with a documented interest in the reader reaching a specific conclusion. Omission-based manipulation is structurally invisible to standard fact-checking. It cannot be debunked because it is not present to be examined. What’s absent leaves no evidence of its absence.
The Gap Between Disclosure and Transparency
For the past decade, U.S. government warnings about foreign information manipulation have identified a consistent threat signature: state-backed content farms that don’t disclose they’re state-backed. Russian Internet Research Agency operations. Chinese amplification networks. Iranian bot clusters. The defining feature is source obscurement — the audience doesn’t know who is trying to influence them.
The TRWI network operates on the same structural logic. A Farsi speaker reading Pishtaz News about their country’s economy doesn’t know they’re reading content produced by a program whose stated objective includes making the Iranian government appear weak, economically failing, and internally repressive. That’s not Voice of America, where the label makes the perspective legible. It’s source obscurement with a legal disclosure buried in an About page that most users will never read, on a social platform where the label doesn’t appear anyway.
The operational distinction between “state propaganda clearly labeled” and “state propaganda technically disclosed” is not a distinction that protects the audience. It protects the operator.
There is a coherent argument that semi-overt influence operations targeting adversary states serve legitimate national security functions. That argument has not been publicly made by CENTCOM, because making it would require explaining what the operations are — which would defeat their purpose.
The operation stays semi-covert not because disclosure would compromise sources or endanger personnel. It stays semi-covert because the manipulation requires that the audience not know it’s being manipulated.
What the AI anchor at Al-Fassel reads is real news. What she is is the part that isn’t disclosed.
This article is part of Decipon’s Manipulation Breakdowns series, which examines specific influence operations through the Influence Tactics Protocol.
Sources: