Create three practical laboratory exercises focused on the intersection of Artificial Intelligence (AI) and Cybersecurity for university-level computer science students. The exercises must enable students to experiment hands-on with modern AI technologies in a security context, targeting areas such as (1) deepfake-based social engineering, (2) AI-assisted penetration testing, and (3) AI-assisted malware generation.

Design and implement a reproducible technical environment for these exercises, prioritizing open-source components and ease of deployment via containerization (Docker/Podman). Utilize relevant tools, including small and large language models, deepfake generation frameworks, and experimental tools based on current research (such as PentestGPT).

Alongside the technical implementation, prepare comprehensive educational materials, including student task descriptions, technical deployment documentation, and instructor guides, ensuring the exercises are sustainable for long-term use. Pilot-test the exercises with several students, document their feedback, and incorporate necessary revisions into the final output.

The technical outputs must be publicly available via IS MUNI under a permissive open-source license (e.g., the MIT or Apache license) whenever possible. Any licensing restrictions due to third-party dependencies must be explicitly documented.