The short version
Key points
- Meta says content that could incite violence or interfere with voting will be removed.
- Other content debunked by independent fact-checkers may be labelled and shown to fewer users.
- The plan involves the Australian Associated Press and Agence France-Presse as fact-checking partners.
- Users will be encouraged to disclose when they share AI-generated images or videos.
- A previous deepfake investment scam was promoted as an advertisement before being taken down.
What Meta says it will do
Meta says it will remove false content that violates its rules, including material that could incite violence or interfere with people’s ability to vote. Content that does not meet the removal threshold but is found to be false by independent fact-checkers may receive a warning label and be downranked in users’ feeds.
The company also says it will identify or label altered and AI-generated content where appropriate. Users are being encouraged to disclose when images or videos have been created using artificial intelligence. Meta’s Australian policy head said the aim is to help people recognise when photorealistic content is synthetic.
Fact-checking and media literacy
The Australian Associated Press and Agence France-Presse are identified as Meta’s fact-checking partners for the initiative. Meta also plans to work with the AAP on a media literacy campaign intended to help Australians assess online content more critically.
The proposed system depends on false claims being identified, checked and acted on quickly. That creates a practical challenge during an election campaign, when a misleading post can be shared widely before a correction or warning appears. Warning labels and reduced distribution may limit later exposure, but they do not necessarily undo the initial spread of a claim.
Why Meta’s record is attracting scepticism
The announcement comes with a history that has made Meta’s promises difficult to assess. During the 2019 election campaign, a false claim about Labor introducing a ‘death tax’ circulated widely on Facebook. Labor called for the claim to be removed, but Facebook declined to take it down at the time, arguing that it was not the platform’s role to act as a censor.
Meta also ended its third-party fact-checking programme in the United States in January 2025 and moved towards a Community Notes-style approach. The company is therefore promoting a fact-checking strategy in Australia that it has recently abandoned in the US, creating questions about whether the Australian approach is a long-term commitment or a response to local political pressure.
The timing is also notable. Australia’s federal election is due by May 2025, and the announcement follows debate about possible laws requiring platforms to address misinformation. The Australian Government later dropped its proposed online misinformation regulation plan after it failed to pass the Senate. Meta’s voluntary measures may therefore also be intended to demonstrate that the company can regulate content without new penalties.
The deepfake problem
Deepfakes present a particularly difficult moderation problem because convincing fake video and audio can be produced quickly and may be hard to distinguish from genuine material. The transcript refers to research involving Australian and Korean researchers in which none of 16 tested detection systems could reliably identify deepfakes in real-world conditions.
A recent Australian example illustrates the risk. A fake current-affairs-style video appeared to show Treasurer Jim Chalmers and business figures endorsing an investment scheme. The video was promoted as an advertisement on Facebook and Instagram, reportedly reaching about 100,000 Australians after scammers spent approximately $7,000. It was removed after public attention, including a warning from Dick Smith, but only after many people had seen it.
A politically motivated deepfake released shortly before polling could create similar difficulties. Even if it were eventually identified and labelled, the correction might arrive after the material had influenced public discussion.
What it means for voters
Effective enforcement could reduce the visibility of false claims, discourage some bad actors and make it easier to identify manipulated media. However, relying on a private platform to decide what users see also gives Meta a significant role in Australia’s political information environment.
Voters still need to assess claims independently. Checking sensational material against reliable news sources, looking for confirmation before sharing and treating unusually scandalous videos with caution can reduce the risk of amplifying false content. The Australian Electoral Commission also maintains a disinformation register to address common electoral myths.
TechManPat’s conclusion
I’m not convinced Meta’s announcement is enough to demonstrate that misinformation and deepfakes will be controlled during the 2025 election. The measures could help if they are enforced quickly and consistently, but Meta’s previous failures, the company’s changing fact-checking strategy and the difficulty of detecting realistic deepfakes justify continued scrutiny. My view is that voters should stay sceptical, report clearly false content and avoid allowing platform algorithms to do the thinking for them.
This knowledge-centre summary is based on the linked TechManPat video and reflects the information available when it was published. Check current pricing, availability and policies before acting.
