
alexdevassy/Machine_Learning_CTF_Challenges
π¦ Open Source Projectalexdevassy
Hands-on CTF challenges focused on exploiting AI models, LLMs, and autonomous AI agents.
This repository serves as a practical training ground for AI security professionals and penetration testers. It covers a wide spectrum of vulnerabilities inherent in modern AI systems, moving beyond traditional software security into the domain of adversarial machine learning. The challenges are structured to demonstrate how attackers can manipulate model inputs, bypass safety filters in LLMs, and exploit logic flaws in AI agent workflows.
Key technical areas covered include adversarial perturbations, prompt injection techniques, data poisoning simulations, and insecure model deployment practices. By engaging with these challenges, users gain hands-on experience with the Python-based tools and methodologies required to identify and mitigate risks in AI-integrated applications. The project is an essential resource for those looking to bridge the gap between traditional cybersecurity and the emerging field of AI safety and security.
π‘Highlights
- ββCovers LLM & AI agent exploits
- ββAdversarial ML attack scenarios
- ββHands-on Python-based challenges
π―For
- ββSecurity Researchers
- ββAI Engineers
- ββPenetration Testers
πLinks
- ββGitHub Repository