Survey Defense Hackathon

Apply

Saturday, February 21, 2026

Remote • One-day event

Artificial intelligence bots can now impersonate real people and complete surveys at scale. This hackathon tests what actually works to stop them.

Hosted by
University of Michigan
&
Princeton University

How the hackathon works

Days 1–6 Prep Week Build your bot or defense
Day 7 Compete Test bots vs defenses

Teams are assigned to one of two roles:

Offense: Build bots that attempt to complete a survey.
Defense: Design safeguards to detect or block bots.

Apply solo or as a team of 2–3.

Why participate?

  • Applied AI, automation, and survey experience
  • Contribute to research on AI and survey methods
  • List on your résumé as Michigan–Princeton Survey Defense Hackathon

Awards

Offense 1st & 2nd place
Defense 1st & 2nd place

Judged on cost-effectiveness. All participants acknowledged in published research.

Details

Format

  • Fully remote
  • Single Saturday event
  • Pre-event prep; no late-night marathons

Tools

  • Any programming language
  • Web automation, AI/LLMs allowed
  • Participants should have basic coding experience

Resources

  • Same small budget for all teams
  • Use any services (e.g., APIs, automation tools); report spending
  • Capped for fair comparison

Ethics

  • Bot code not released
  • Defense ideas may be used in research (with credit)

Apply

Short intake survey: background, interests, role preference.

Start Application

FAQ

Who can participate?

Undergrad/grad students from any university.

Do I need CS experience?

Yes, some level of basic coding experience.

Can I choose my role?

Indicate preference; final assignment balances teams.

Can I apply with friends?

Yes, as a team of 2–3 or individually.

How much time?

Prep beforehand + full Saturday (ends early evening).

What tools allowed?

Scripting, automation, AI/LLMs. Rules provided in advance.

Will code be shared?

Defense ideas may be refined for research. Bot code stays private.

Good for my résumé?

Yes—concrete AI + security + data quality experience.