Hack the Border 2025 — Adversarial Thinking & Cybersecurity with AI

Theme

Defend the Future — Adversarial Thinking & Cybersecurity with AI — build or educate for people-first defenses against modern threats (phishing, social engineering, deepfakes/voice clones, data mishandling) using AI as a tool for defense and a lens to reason about risk. Projects should be transparent, safe, useful, and privacy-respecting, teaching communities to anticipate, test, and counter attacks with clear, adoptable practices.

Project (Optional)

Over ~6 weeks, create a focused prototype (web/app/bot) or an educational piece (video/carousel) that solves a real local need; submit on Devpost one week before the event with a mandatory video.

Scoring

Final score = 50% Project (judged during the week before) + 50% Villages.

Villages (Day 2)

Earn points by participating in CTF, Secure Coding, and Other Activities—beginner-friendly; partial progress still counts.

Event Flow

Day 1 = Registration + Industry Panel Night • Day 2 = Villages, gallery-style showcase (we’ll play your videos), lightning demos, and awards.

Track 1 — Adversarial Thinking

Help people spot risks and choose simple defenses—think like attacker and defender (ethically).

Example deliverables:

• Scenario Generator (attack/defend): Short caselets (AI phish, voice clone, social engineering, data mishandling) with reflection questions + facilitator notes. MVP: 5 scenarios + PDF export. Stretch: deck export and difficulty levels.

• Prompt Jailbreak Lab (beginner-safe): Demo a risky prompt → show the safer rewrite and guardrails; explain why it works. MVP: 3 jailbreaks + fixes. Stretch: side-by-side output viewer + “copy safe prompt.”

• Attack-Tree → Blue-Plan Builder: Sketch threats for a chosen context; tool suggests countermeasures and a 15-minute tabletop drill sheet. MVP: drag-and-drop nodes + printable drill. Stretch: library of common threats/controls.

Notes: Keep ethics front-and-center; avoid real sensitive data; include a lightweight README (what/how/limits) and the video showing threat → defense → how to use it.

Track 2 — Community Cybersecurity

Make everyday digital life safer for people and small orgs—simple, respectful, adoptable.

Example deliverables:

• Cyber Hygiene Coach (web app): Quick self-audit → prioritized fixes (passwords/MFA/updates/backups) with how-to guides. MVP: checklist + links. Stretch: reminders, progress tracking, printable report.

• Phishing Drill-in-a-Box: Click-through simulations + debrief pages and talk tracks for clubs/nonprofits (no real PII). MVP: 3 scenarios + debrief notes. Stretch: results dashboard, auto-generated feedback email.

• Public Computer/Kiosk Hardening Wizard: “Before you log off” checklist (clear cache/downloads/clipboard) + admin setup tips; printable station placards. MVP: single page + PDF. Stretch: timer and end-of-session prompts.

Optional education asset for any of the above: 2–5 min video or one-pager “Top 10 fixes in 10 minutes” (EN/ES).

Track 3 — Prompt Engineering

Build reusable, responsible prompting foundations for everyone.

Example deliverables:

• Prompt Safety Checker (web app): Paste a prompt → flag privacy/bias/policy risks, show safer rewrites. MVP: rule-based checks + red/yellow/green + 3 rewrites. Stretch: domain presets; revision log.

• Prompt Pattern Library: Taggable gallery of proven patterns (Tutor, Critic, SQLizer), with “when to use,” pitfalls, copy-to-clipboard. Stretch: side-by-side “test this pattern” sandbox.

• Prompt Robustness Tester: Auto-perturb prompts (typos/synonyms/order) and show stability metrics + suggested guardrails.

• Persona Prompt Composer: Wizard to generate role prompts (translator, safety reviewer) with tone, constraints, refuse rules, and a disclosure line. Stretch: save/share presets.

Optional education asset for any of the above: 1–3 page “Prompt Do/Don’t” cards or a 2–5 min micro-lesson.

Track 4 — AI Study & Helper Bots

Design helpful, privacy-aware assistants for learning and everyday community services.

Example deliverables:

Learning & study

• Syllabus-to-Study-Plan: Upload a syllabus → weekly plan, practice prompts, spaced-repetition reminders. Stretch: calendar sync, adaptive difficulty.

• “Explain Like I’m New Here” Tutor: Intake prior knowledge → stepwise explanation → mini-quiz with feedback; bilingual toggle.

• Study Group Mode: Shared room where the bot rotates prompts, keeps time, assigns roles (explainer/checker).

Community & services

• Small-Business Policy Buddy: Generates plain-language policies (passwords/updates/backups) + rollout plan and to-do list.

• Event Safety & Comms Assistant: Drafts safety briefings, signage text, lost-and-found workflow, emergency SMS templates; PDF export.

• Library/Kiosk Guardian: End-of-session checklist (clear cache/downloads/clipboard) + timers/reminders; optional admin analytics.

Notes: For any bot, include a lightweight README (what/how/limits), a privacy note (no real PII), and a mandatory video.

Deliverable guidance (applies to all tracks)

• Build (prototype): working slice (web/app/bot/extension) + README (what it does, how to try, limits, what’s next).

• Educate (media): 2–5 min video or 5–8 tile carousel + sources + “do this next” checklist.

• Mandatory video: overview for all submissions (used in showcase).

• Impact & Relevance
  Rates how well the project addresses a real, local need and benefits a clear audience. Strong entries show community context, credible use cases, and a path to adoption.
• Clarity & UX
  Evaluates how easy it is to understand and use. We look for plain language, clean flow, and inclusive design (accessibility and, where helpful, bilingual support).
• Originality
  Measures freshness of the idea or framing and constructive reuse of tools. Credit prior work, but add a clear value-add or novel angle.
• Functionality / Instructional Quality
  For Build entries: does something run and demonstrate the core workflow with basic documentation and guardrails? For Educate entries: is the content accurate, well-sourced, and effective at teaching a practical takeaway?
• Presentation
  Assesses storytelling and time discipline. Great demos make the problem, solution, impact, and next steps obvious—without going over time.
• Effort & Teamwork
  Considers scope realism, visible collaboration, and iteration. Show roles, progress, and what you learned or changed along the way.