Personal Security
A tool that protects individuals — their data, identity, finances, family, or wellbeing.
What this prompt is really asking.
AI is making scams cheaper, faster, and more convincing. Voice cloning, phishing-as-a-service, deepfaked verification calls, AI-generated tax fraud — all of it is moving downmarket fast.
Build something that protects an individual — their money, their identity, their family, their data, or their wellbeing. Specific. Useful. Defensive. The bar is: a real person can use this when something is going wrong, today.
The defenders are losing ground. Banks deploy AI fraud detection; consumers don’t. Most of the people MLT serves are first lines of defense for their parents, elders, and communities. Tools designed for them are scarce.
Concrete starting points.
Pick one. Or invent one of your own — these are starting points, not assignments.
- 1
Scam detectors for text and voice
Paste a suspicious text, voicemail, or screenshot. Get a confidence score, what the scammer wants, and the exact next step.
- 2
Privacy audits for non-engineers
Walk a user through what their phone, browser, and main apps know about them — and how to turn it off in plain language.
- 3
Estate and emergency-doc walkthroughs
Wills, healthcare proxies, beneficiary forms — generate the right documents for the user’s state and remind them what to update yearly.
- 4
Elder-protection co-pilots
Detect anomalous bank or phone activity for an aging family member. Send a plain-language alert to the designated trusted contact.
- 5
Identity-theft response agents
Once a breach happens, walk the user through the exact 10 steps — credit freeze, IRS PIN, account locks — and track which are done.
How judges will know it worked.
A real person can use it the moment something is going wrong, not as a study tool months ahead.
Sensitive inputs do not leak. The data model is documented and conservative.
There is a path to scale: paying users, partnership with a financial institution or community org, or foundation funding.
Where most teams will start.
Use anything you like — these are the ones we’d reach for first. Full quickstarts on the resources page.
Claude Code
Reliable for backend logic with sensitive data — good defaults around safety and review.
Replit
AI Agent + hosting for tools that need a phone-friendly UI fast.
Google AI Studio
Free Gemini Flash for text classification (scam vs. not), structured extraction from documents.
Mentors paired at kickoff. Office hours run daily during build week.
Still deciding? Read the other two.
Pick "Personal Security" at signup.
You can change your mind once before kickoff. After that, your prompt is locked.