Foreign 'AI slopaganda' network deepfakes dozens of Australian politicians | Cloud based enterprise resource planning
By Gammatek ISPL, Industrial Systems & Compliance Analyst at Gammatek ISPL
Last updated: September 2026 | 14 min read
Author block: Gammatek ISPL covers cybersecurity and compliance trends affecting manufacturing, chemical, and pharma plants at Gammatek ISPL. This piece draws on public reporting from ABC News Australia and Reset Tech's published research, alongside Gammatek's direct experience advising clients on network security architecture.
Why This Matters to You
A foreign-run network of more than 200 Facebook pages has spent the past eight months churning out AI-generated deepfakes of Australia's Prime Minister, opposition leader, and other political figures — and roughly one in ten of those pages was earning money directly from the platform while doing it. If this sounds like a story purely about politics, look again: the exact same production process — cheap AI generation tools, weak platform detection, and a monetization system that rewards engagement over accuracy — is already being pointed at businesses, not just politicians. If your organization handles wire transfers, vendor payments, or executive approvals over video or voice calls, this story isn't background noise. It's a preview of a fraud technique that has already hit companies, and the defenses most businesses have in place were not built with this threat in mind.
What Actually Happened
According to reporting from ABC News Australia, research organization Reset Tech tracked more than 200 Facebook pages impersonating Prime Minister Anthony Albanese, Opposition Leader Angus Taylor, One Nation Senator Pauline Hanson, and numerous other political figures over an eight-month period. The pages were linked back to an influencer business based in Sri Lanka, and researchers described the operation as a "foreign influence" network rather than traditional state-sponsored interference — the motive appeared to be commercial, not purely political.
Roughly 10 percent of the network's pages had been monetized through Facebook's own creator payment programs, meaning the people running the network could profit directly from the platform for producing and distributing the deepfake content. Reset Tech's report found that platform enforcement was inconsistent at best: only about one in three of the flagged posts had received an AI-generated content label, and only about one in 33 had actually been removed, despite the content violating Facebook's stated policies on inauthentic behavior, hate speech, and AI-generated deepfakes.
Researchers described this as part of a broader pattern of "AI slopaganda" — a term used to describe low-cost, high-volume AI-generated propaganda content — operating at what one researcher called an "industrial scale" against political systems generally, not just in Australia.
This is not an isolated incident. Separate reporting has documented earlier cases of Australian officials, including Finance Minister Katy Gallagher and Foreign Minister Penny Wong, being depicted in AI-generated deepfake videos used in investment scams as early as 2024.
Why This Keeps Happening: The Incentive Structure
The most important detail in this story isn't the technology — deepfake generation tools have been accessible for years. It's the incentive structure that makes running a network like this worthwhile. Three factors combine to make this attractive to bad actors:
Generation cost has collapsed. Producing a convincing AI image or short video clip now takes minutes and costs a fraction of what it did even two years ago.
Platform monetization rewards engagement, not accuracy. A page earning ad revenue or creator payouts based on views has a direct financial incentive to produce whatever content generates the most engagement — and divisive political deepfakes reliably do.
Detection and enforcement lag behind production. Reset Tech's own researchers noted they were able to identify these networks with a small research team, which raises an obvious question about why platforms with far larger resources have detected and removed so little of it.
This same incentive structure — low cost to produce, weak detection, and a payoff for the person doing it — is exactly what's driving deepfake fraud against businesses, just with a different payoff (fraudulent payments instead of ad revenue).
The Business Version of This Threat
Political Deepfake Networks | Business Deepfake Fraud | |
Goal | Engagement, ad revenue, political influence | Direct financial theft, fraudulent authorization |
Method | AI-generated images/video posted at scale on social platforms | AI-generated voice or video used in real-time calls or pre-recorded messages |
Target | Public opinion, platform algorithms | Finance/accounts payable staff, executives, vendors |
Detection difficulty | Platform-level content moderation, often reactive | Person-to-person, often happening in real time with no automated check |
Documented cases | Reset Tech's 200+ page network (2026); earlier scam videos using Australian ministers (2024) | Multiple reported cases of AI voice cloning used to authorize fraudulent wire transfers globally since 2023 |
Deepfake-enabled business fraud isn't hypothetical. Since 2023, there have been multiple widely reported cases of finance staff at companies transferring large sums after receiving what appeared to be a video call or voice instruction from a senior executive — later confirmed to be an AI-generated impersonation. Manufacturing and industrial companies are not exempt from this risk; large one-time payments to vendors, contractors, and equipment suppliers are a normal part of plant operations, which makes finance and procurement staff at these companies a realistic target for the same fraud pattern.
An Implementation Consideration for Manufacturing and Industrial Companies
If a foreign network with modest resources can convincingly impersonate a sitting Prime Minister at scale, a well-resourced fraud operation can convincingly impersonate a plant manager, CFO, or vendor contact in a single targeted call — and single targeted calls require far less content than a 200-page social media operation.
Practical considerations worth reviewing internally:
Verbal or video authorization alone is no longer a sufficient control for high-value payments. Any payment or contract approval process that relies solely on "I heard/saw the executive approve this on a call" should have a secondary, out-of-band verification step — a callback to a known number, a separate confirmation channel, or a pre-agreed verification phrase.
Staff training needs a specific deepfake scenario, not just general phishing awareness. Most security awareness training still focuses on suspicious emails and links. Few programs specifically train finance and procurement staff to recognize the signs of a real-time voice or video impersonation attempt.
Vendor and supplier communication channels deserve the same scrutiny. A deepfaked "vendor" requesting updated bank details is a documented fraud pattern that predates AI deepfakes (traditional business email compromise) and is now easier to make convincing with synthetic voice or video.
Incident response plans should explicitly cover deepfake-enabled fraud as a distinct scenario from traditional phishing or ransomware, since the detection signs and immediate response steps differ.
What Platforms and Regulators Are (and Aren't) Doing
Reset Tech's findings point to a broader accountability gap: platform enforcement of existing policies against AI deepfakes and inauthentic behavior networks is inconsistent, even when the violations are documented and reported. Australia has been developing AI governance frameworks and has debated truth-in-political-advertising laws at the state level, though comprehensive federal deepfake-specific legislation remains a work in progress, consistent with the broader global pattern of regulation lagging behind the technology's deployment.
For businesses, this means waiting for regulation or platform enforcement to solve this problem is not a realistic near-term strategy. The practical response has to happen at the organizational level — through verification processes, staff training, and technical controls — rather than assuming external enforcement will close the gap.
Where This Is Headed
The Australian political deepfake network is a visible, well-documented case precisely because researchers were looking for it and journalists covered it. Business-targeted deepfake fraud is far less visible by design — a successful fraudulent wire transfer doesn't generate a news story, it generates a quiet financial loss and, often, an internal decision not to publicize what happened. The actual scale of business deepfake fraud is likely underreported relative to the political version of this same technique, precisely because there's less incentive to talk about it publicly.
The lesson from watching a political version of this play out at scale is that the underlying weaknesses — cheap generation, weak detection, and processes that assume a voice or face on a call is sufficient proof of identity — apply directly to how most companies still handle high-value approvals today.
How Gammatek Approaches This
Plant security architecture conversations increasingly need to include this scenario alongside more familiar threats like ransomware and network intrusion. Reviewing where your organization still relies on a phone call or video call as the sole authorization step for high-value actions is a reasonable starting point, and it connects directly to the broader network security planning already covered in our related posts.
See how Gammatek's compliance and security advisory work helps manufacturers close verification gaps like this →https://www.gammateksolutions.com/ https://www.gammateksolutions.com/post/fortinet-cyber-security-pricing-2026-firewall-cost-guide
Comments