AI as a Weapon in the Cyber World

• Strategy: cognitive warfare, tempo, attribution

• Authenticity: C2PA/metadata, deepfake triage

• Threat mapping: poisoning, backdoors, injections, exfiltration

• Pipeline hardening: RAG/agents with intake checks, canaries, policies

• Detection & robustness: misuse detectors, anomaly scoring, ART/TextAttack/Foolbox

• Ops & governance: KPIs, monitoring, incident response, AAR

📅 Day 1 – Strategy, OSINT & Authenticity

• Weaponization: capability × intent × doctrine; supply-chain risks
• Authenticity stack: C2PA, watermark limits, provenance graphs
• Hands-on: Provenance verifier (C2PA/EXIF + heuristics), benign OSINT graphing, deepfake triage notebook
• Free tools: SpiderFoot, theHarvester, recon-ng, exiftool, C2PA CLI, InVID-WeVerify, Python/Jupyter/OpenCV/librosa

📅 Day 2 – Adversarial AI & Model Security

• Threat surface: poisoning, backdoors, prompt/indirect injection, exfiltration
• Defense-in-depth; purple-team mapping to detections/controls
• Hands-on: Policy-driven two-pass RAG (local LLM), robustness demo (ART/TextAttack), telemetry anomaly scoring
• Free tools: IBM ART, TextAttack, Foolbox, scikit-learn, PyTorch, LangChain, LlamaIndex, Ollama/llama.cpp/GPT4All, promptfoo, Guardrails

📅 Day 3 – Detection, Dashboards & Policies

• Hands-on:
  • LLM misuse detector + Streamlit mini-dashboard
  • Provenance verifier (batch CLI) with policy actions
  • RAG intake hardening (hash/MIME checks, denylists, canaries)
  • Optional: autoencoder vs IsolationForest comparison
• Free tools: CICIDS-2017/UNSW-NB15, scikit-learn, PyTorch, Streamlit/Grafana, Zeek (optional)

• Strategy lens: cognitive warfare, tempo, attribution

• Authenticity & provenance (C2PA), deepfake triage

• Defense-in-depth: data → model → agent → ops

• Hardened RAG/agents with policy controls & canaries

• Adversarial threats handled: poisoning/backdoors/injections/exfil

• Operational detectors & robustness checks (ART/TextAttack/Foolbox)

• PhD scholars, postgraduates, and senior undergraduates in AI/CS/Cybersecurity

• Professors, researchers, and lab leads working on AI security or policy

• Security architects, red/purple-teamers, SOC/DFIR analysts, and threat researchers

• ML/AI engineers, data scientists, and platform/MLOps engineers in safety-critical domains

• Government/defense, CERTs, and critical-infrastructure practitioners

• Product/Policy leaders responsible for safe AI deployment and governance