Bot Farms, Fake Accounts, and Deep-Pocketed Donors: Inside One Nation’s Online Machine

ByPriscilla Li

One Nation’s 2026 online surge is best explained by four distinct layers: coordinated real supporters on Facebook, foreign AI-generated fake-news networks, engagement-driven recommendation feeds, and disclosed donor money that helps sustain the operation. The reporting does not point to one master switch; it documents separate systems that reinforce each other, from domestic supporters pushing posts into the feed to Vietnam-run pages publishing fabricated pro-Hanson stories and donation disclosures showing money flowing to Pauline Hanson’s One Nation.

The useful distinction is that not all of this is “bots.” Some of it was real people acting in coordination, some of it was synthetic foreign spam, and some of it was the ordinary logic of ad-funded feeds: if a post gets reactions, comments, and shares, the ranking system treats it like a live wire.

Coordinated human accounts on Facebook

The domestic part of the operation appears to have been built around real Facebook users posting and engaging in concert, not fleets of automated accounts. Crikey’s reporting on One Nation’s Facebook-centered rise in the 2026 Farrer by-election describes a communication strategy designed to bypass traditional media and lean hard on Facebook distribution.

That matters because normal bot-detection systems are mostly built to catch the wrong thing here. Platforms are relatively good at looking for signals such as suspicious account creation patterns, machine-like posting cadence, or shared technical fingerprints. A network of genuine supporters using their own longstanding accounts can look perfectly normal at the account level while still behaving like a swarm at the content level. It is the difference between a script and a flash mob.

That is why this should not be lumped together with familiar debates about bot-like or fake accounts on X or broader estimates of bots on X. Those debates are about fake or automated accounts. The One Nation pattern described in accessible reporting is different: real users manufacturing momentum by concentrating attention at the same time and place.

Crikey’s paywalled reporting limits how far the public record goes on specific operational details, so the safe conclusion is the broad one: a Facebook-first domestic supporter network can trigger recommendation systems without tripping classic bot alarms, because the accounts themselves are authentic people.

Vietnam-run AI fake-news pages

A second layer was much more literal disinformation. ABC News Verify reported on 11 March 2026 that foreign Facebook accounts and pages were using AI-generated and genuine images of Pauline Hanson alongside fabricated stories to manipulate Australian audiences. The report said some of the pages were administered from Vietnam, and that they pushed false claims designed to flatter Hanson and inflame her opponents.

One example ABC documented was a fabricated story claiming Hanson was suing the prime minister in a US court for defamation. The story used invented quotes and did not describe any real case, but it was packaged to look like political news and fed into Facebook anyway.

AFP documented the same system, describing “industrial” clickbait disinformation tied to Vietnam-managed Facebook pages publishing AI-generated articles and false claims about Australian politics, including hoax stories about Hanson. AAP also reported that Vietnam-linked pages were masquerading as Australian news outlets while publishing inaccurate AI-generated political content.

These pages were not just partisan spin. They were fabrication mills: fake outlet branding, invented legal actions, synthetic or misleading imagery, and engagement bait tuned for distribution.

The sharp distinction is worth keeping. The domestic supporter layer was about coordinated humans. The foreign disinformation layer was about fake content factories. Both can end up producing the same platform signal, engagement, but they are different systems with different failure modes.

Donation records and disclosed backers

The money layer is less dramatic than the fake-news pages, but it is easier to document. The Saturday Paper reported that Barnaby Joyce said Gina Rinehart was funding One Nation and described support shifting away from Coalition parties. Crikey separately reported on large donations by executives inside Rinehart’s Hancock network and disclosure-linked questions around a plane gift.

The disclosed-record view comes from the Australian Electoral Commission’s Transparency Register and aggregators built from it. DonationWatch’s One Nation page lists disclosed donor transactions for Pauline Hanson’s One Nation by date and amount, drawing from AEC data, while its Australia overview notes that the underlying records can miss or misclassify entries because of source-data errors and disclosure thresholds.

That limitation matters, but the disclosed pattern is still useful. Public reporting and disclosure-linked databases show named money coming from wealthy backers, companies, and business-linked donors, not just small-dollar grassroots support. The Guardian’s summary of 2024-25 Australian political donation disclosures also noted One Nation’s receipts in the context of broader right-wing funding flows.

Published donations data can tell you who gave disclosed money and when. It cannot, by itself, prove any explicit quid pro quo over policy or content strategy. It also understates total funding, because the AEC publishes annual returns on the first business day in February for the prior financial year and election returns 24 weeks after polling day, so the public record arrives late, and some receipts can sit below disclosure thresholds.

The feed amplifies whatever wins attention

The fourth layer is the recommendation system itself. Facebook, TikTok, and YouTube are ad-funded platforms. Their ranking systems are built to maximize time, reactions, shares, comments, and repeat visits because that attention is what gets sold to advertisers.

That does not prove any platform set out to boost One Nation specifically. It does explain why high-arousal political content tends to travel farther than calmer or more corrective material, especially when someone is already priming the system with early engagement. Whether the spark comes from coordinated supporters or AI-generated fake pages, the feed mostly sees the same thing: people are interacting, so keep showing it.

Here the mechanism is simpler than the politics. Recommendation systems are giant relevance machines with a crude proxy for value. If outrage, tribal identity, and scandal language produce more interaction than measured reporting, the machine learns the wrong lesson very quickly.

“One Nation’s 2026 online surge is best explained by four distinct layers: coordinated real supporters on Facebook, foreign AI-generated fake-news networks, engagement-driven recommendation feeds, and disclosed donor money that helps sustain the operation.”

That is also why “just remove the bots” is too small an answer. Even if platforms got much better at detecting synthetic accounts, they would still face organized human participation and ranking systems that reward whatever looks lively.



What the evidence supports

The strongest sourced answer is not that One Nation’s surge came from one hidden trick. It is that four documented systems stacked together.

The domestic Facebook network explains how a party can look bigger online than traditional media coverage would suggest. The Vietnam-run AI pages explain how fabricated pro-Hanson stories entered the same information stream. The ad-funded feed explains why both kinds of engagement can be amplified without any platform operator needing to choose a party. And the donations record shows there is disclosed money behind the broader machine.

The next major update to the public record will come from future AEC disclosure releases and election returns, which are published on a fixed timetable rather than in real time.

Key Takeaways

  • One Nation’s 2026 online surge is best explained by four layers: coordinated human supporters, foreign AI fakes, engagement-optimized feeds, and donor money.
  • ABC News Verify, AFP, and AAP each documented Vietnam-linked Facebook pages publishing fabricated pro-Hanson political content.
  • The domestic Facebook layer appears to have relied on real coordinated supporters, which is why normal bot-detection systems would be less effective.
  • DonationWatch, built from AEC disclosures, shows disclosed money flowing to One Nation, though the public record is delayed and incomplete in real time.
  • The claim that feeds amplified this content is a structural inference about engagement-based ranking, not direct proof that any platform intended to favor One Nation.

Further Reading