Inspiration

Security teams are burnt out trying to guard systems, while attackers have it easy. Today, malicious hackers don't need to spend weeks manually picking apart software–they can just plug an AI directly into reverse engineering tools like Ghidra or IDA to clone, steal, or find vulnerabilities in code in seconds. We realized that defending software manually is a losing battle, so we asked: What if we turned the AI's own mechanics against it? Instead of fighting AI with more firewalls, we wanted to build a shield that tricks the AI into breaking its own analysis.

What it does

It is a pre-build tool that protects your private code from automated AI theft.

How we built it

Using word to store our research, and GitHub to share the code with each other. We used Ibis Paint to do the Logo and Mascot (Chip).

Challenges we ran into

Coding a prompt injection payload that actually worked in misleading the ai.

Accomplishments that we're proud of

Getting a creative idea that can help current cybersecurity issues.

What we learned

A rough draft is better than having nothing, because you can build the draft into something better.

What's next for C.A.R.E

We plan to build C.A.R.E. directly into modern development setups, like GitHub Actions, so code is automatically protected right before it is published. Expanding our scope to parsing and C++ so that it gets protected too.

Built With

Share this project:

Updates

Submission history