odyss3y

AI red-teaming and safety evaluation across LLMs, diffusion models, and agentic systems. Focus areas: refusal behavior, alignment drift, bias, provenance, accountability, and narrative agency preservation. Cybersecurity/DFIR background.