Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a six-attribute taxonomy for profiling threat actors in the context of pre-release risk management for open-weight AI models. By grounding the taxonomy in existing literature from terrorism, biosecurity, and cybersecurity, the authors argue that explicit adversary characterization is essential for creating interpretable and comparable risk evaluations. The key finding is that this structured approach can significantly enhance the decision-making process for developers by providing a common language to articulate and assess adversary assumptions prior to model release.
Explicitly characterizing adversaries could transform how we evaluate risks associated with releasing open-weight AI models.
Pre-release risk management for frontier AI misuse risks routinely leaves threat actor assumptions implicit, inconsistently specified, or ungrounded. This capstone argues that explicit adversary characterization should be regarded as a prerequisite for evaluations that are interpretable, comparable, and faithful to the risks they target. We propose a six-attribute taxonomy (covering technical sophistication, prior domain knowledge, organizational capacity, operational infrastructure, financial capacity, and time horizon) with empirically grounded tiers derived from existing terrorism, biosecurity, and cybersecurity literature. The taxonomy is designed to function as research infrastructure: a common language for pre-specifying adversary assumptions before evaluations are conducted, analogous to pre-analysis plans for randomized controlled trials (RCTs) in medicine and economics. Its application is particularly urgent for open-weight model developers, for whom release decisions are irreversible and must anticipate adversarial reasoning.