There is a particular kind of lesson that lands harder than any lecture, and a group of Australian students appear to have just lived through one. In a classroom exercise designed to teach how modern information warfare actually works, students were handed the tools of the trade, including automated AI bots and the deliberately provocative content known online as “rage bait”, and set loose on a fictional election. The result, as reported by the ABC, was a sobering demonstration of how little effort it now takes to tilt a public conversation.
The premise is deceptively simple. Rather than warning young people in the abstract about misinformation, the exercise puts them on the other side of the keyboard. Students become the manipulators, building campaigns around outrage, running bot accounts that amplify a message, and watching in real time how sentiment around a candidate can be nudged, hardened or inflamed. By the end, the lesson is not that disinformation exists, which everyone already knows, but how mechanically straightforward it has become to produce and distribute it at scale.
Why rage bait works so well
“Rage bait” is the online cousin of clickbait, except the currency is anger rather than curiosity. Content is engineered to provoke an emotional reaction strong enough that people share it, argue about it and spread it further, often without pausing to check whether it is true. Social platforms, which reward engagement above almost everything else, tend to push this material harder precisely because it keeps people scrolling and commenting. Layer generative AI on top, and a single student can now spin up dozens of plausible-sounding posts, fake personas and talking points in minutes, work that would once have required a well-resourced team.
That combination is what makes the classroom version so instructive. The AI does the heavy lifting on volume and variety, generating endless variations of a message so it does not look coordinated. The rage bait supplies the emotional hook that makes people pass it along. The bots provide the illusion of consensus, making a fringe view look like a groundswell. Put together in a fictional campaign with no real-world stakes, students can see the machinery operating without doing any actual harm, which is rather the point.
A teaching tool, or a training ground?
Educators who favour this kind of hands-on approach argue that inoculation beats warning. The theory, well established in media literacy research, is that people become far more resistant to manipulation once they have seen how it is built from the inside. Teaching students to recognise the fingerprints of a coordinated campaign, the suspiciously uniform phrasing, the accounts that only ever post about one topic, the emotional escalation that outruns the facts, is more durable than simply telling them to be sceptical. If you have run the con yourself, the thinking goes, you are harder to con.
Not everyone is comfortable with the method. Critics of exercises that hand real manipulation tools to learners worry about normalising the behaviour, or about teaching techniques that could be repurposed outside the classroom. There is a reasonable question about where the line sits between education and a how-to guide, particularly when the AI tools involved are freely available and improving by the month. Others point out that a fictional election is a safe sandbox, but the same tactics deployed against a real candidate, a real referendum or a real community would do genuine damage, and no classroom debrief undoes that in the wild.
Both views can be true at once. The exercise is valuable precisely because the threat is real, and the threat is what makes some observers nervous about rehearsing it. That tension is unlikely to be resolved neatly, and it mirrors a much larger debate playing out across education and public policy about how openly societies should teach the mechanics of the technologies that can be turned against them.
What it means for Australia
For Australia, this is not a hypothetical concern dressed up as a school project. The country has a federal election cycle, an Australian Electoral Commission that already fields complaints about misleading content, and a population that gets a growing share of its political information from social feeds rather than mastheads. The Australian Communications and Media Authority and successive governments have wrestled with how to counter coordinated inauthentic behaviour without trampling legitimate speech, and the arrival of cheap generative AI has sharpened that dilemma considerably. A tactic that a student can master in an afternoon is, by definition, within reach of anyone who wants to use it.
There is also a distinctly Australian angle in who is doing the teaching. Universities and research bodies here, including the CSIRO’s digital and data science arms, have invested in understanding online influence operations and in building the public’s resilience to them. Framing disinformation as a skills-and-literacy problem rather than purely a platform-moderation problem fits the local instinct that defending democracy is a shared civic job, not something to be outsourced entirely to the companies that run the feeds. The counterpoint, familiar to anyone following the sovereign AI debate in Canberra, is that literacy alone cannot keep pace with tools that get cheaper and more convincing every quarter.
What’s next
The immediate value of the exercise is that it travels. A simulation that works in one classroom can be adapted for others, and the lessons scale to workplaces, newsrooms and community groups that increasingly need to spot manufactured outrage in their own channels. Expect more Australian institutions to experiment with this inoculation style of teaching, and expect the debate about its ethics to follow close behind.
The harder question sits above the classroom. If a student with a laptop and off-the-shelf AI can sway a pretend poll, the defences around real elections need to be robust enough to assume bad actors already have the same capabilities. That means better detection of coordinated behaviour, clearer rules on AI-generated political content, and a public that has, ideally, already seen the trick performed. The students who ran the bots this term may turn out to be among the better-prepared voters Australia has, which is a strange kind of reassurance, and a warning about the rest of us.
Sources: ABC News.
