From the Student Success edition of November 1, 2026
UNSW students build AI bots to sway a simulated election in a month-long competition meant to teach them to spot AI manipulation
University of New South Wales; reported by ABC News (Australia) · Higher education computing teaching and AI literacy · Australia (Sydney)
- Publisher
- ABC News (Australia)
- Original publication
- ABC News, 31 October 2026
- Source retrieved
- 2026-11-02
What happened
ABC News reported on 31 October 2026 that UNSW researchers run a month-long competition in which student teams build AI bots to post, like, share and reply on a purpose-built simulated platform, Legit Social, to sway AI-simulated citizens in a fictional referendum. Teams score for attention, changing views and correctly flagging rival bots. A UNSW senior lecturer says the aim is for students to learn attacker and defender techniques in a safe environment, and that researchers use the game to study AI interference. A preliminary paper on last year's competition found some participants turned to high-volume posting because the scoring rewarded engagement. A student moderator said it helped her recognise bot-like posts. No learning outcome data are reported.
Why it matters
An example of experiential AI literacy teaching in higher education, with an early design lesson: scoring rules shape student behaviour. Limited relevance beyond computing courses.
Evidence and measured results
News feature with lecturer and student interviews and reference to a preliminary paper; no evaluation.
Limitations and uncertainty
Self-reported benefit from one student; preliminary paper not reviewed; election context dominates the story.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-02; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Programs want students to understand AI manipulation and agent behaviour, but lecture-based AI literacy is abstract.
- Stakeholders
- Head of computing or media school
- Course convenors
- Research ethics office
- Teaching and learning centre
- Discovery questions
- Do any courses teach AI threats hands-on?
- Is there a safe environment for students to build agents?
- How would you assess detection skills?
- Value hypothesis
- Interpretation: a contained simulation can give students practical experience of AI manipulation and detection.
- Potential engagement
- Design of one sandboxed simulation exercise with scoring and a pre and post skills assessment.
- Unsupported claims
- Improves misinformation detection
- Transfers to non-computing students
- Reduces real-world manipulation
Pre-sales engineering
Role takeaway
- Fit
- Sandboxed multi-agent simulation for teaching and research.
- Architecture and integration
- Purpose-built platform hosting simulated citizens and student agents, with logging, scoring and moderation.
- Prerequisites
- Model access and budget
- Contained hosting
- Logging and scoring engine
- Constraints
- Model costs over a month-long run
- Scoring can be gamed
- No outcome data
- Security
- Keep agents confined to the sandbox; no connection to real platforms or users.
- Proof of value
- Bot-detection accuracy before and after
- Check that scoring does not reward high-volume posting
- Zero activity outside the sandbox
Delivery
Role takeaway
- Work
- Run a contained AI agent simulation exercise with rules, scoring and assessment of learning.
- Dependencies
- Platform maintenance
- Ethics approval where data are studied
- Instructor time for moderation
- Ownership
- Teaching school course convenors, with research ethics oversight.
- Skills and adoption
- Students need briefing on rules and reflection on the ethics of the techniques they practise.
- Governance checkpoints
- Rules of engagement approved
- Ethics review for research use
- Post-competition review of scoring
- Acceptance criteria
- Proposed: measured gain in bot-detection accuracy
- Proposed: no out-of-sandbox activity
- Proposed: scoring revised if volume posting dominates
- Risks
- Practising manipulation without reflection
- Gamed scoring
- Consent issues for research use
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
A sandboxed simulation platform lets students build agents without touching real platforms.
Governance
Who approves, reviews and stays accountable for outcomes?
Offensive-technique teaching needs a contained environment and clear rules, as UNSW describes.
Security and privacy
What data, permissions and controls need testing?
The sandbox avoids real users and real platforms; no other data issues reported.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Builds skills relevant to trust and safety and cybersecurity work; not measured.
Procurement
What should contracts, pricing and exit terms secure?
Limited relevance.
Operating model
Which teams own the service once it runs?
Teaching and research share the same platform, which needs maintenance and ethics oversight.
Publication history
- 2026-11-01Student Success · Issue 554 resources
Stable resource ID: abc-unsw-kingstonia-simulated-election-ai-bot-competition-students-20261031