AI Security

We study how machine learning and large language models can strengthen security — and how they fail. Work spans byte-level models, agentic AI systems, and representation learning for security tasks.

AI Safety Discourse

Alongside the technical work, Abanisenioluwa is a graduate research fellow at BASE Research (from October 2026). The project uses computational text analysis to study how the AI safety field defines and prioritizes risk, whose perspectives shape the discussion, and which voices and countries are underrepresented. It compares catastrophic-risk framings, including loss of control, with present-day concerns such as bias, misuse and reliability, using a corpus of academic papers, community discussions, safety-organization publications and summit documents. The analysis combines embeddings, topic modeling and LLM-assisted classification checked against human-coded examples, and tracks changes in risk framing and citation patterns over time. The aim is to surface blind spots in the conversations that shape AI research priorities, funding and governance.