This project advances the Functional Intentionality Test (FIT), a developer- and regulator-facing framework that reconceptualizes AI “intentionality” as a functional property of behavior rather than a mental or metaphysical state. In law, intent is inferred from conduct, patterns, and context—not from direct access to mental states. FIT extends this logic to machine learning systems by translating five legally grounded dimensions of intentionality—purpose, foresight, volition, temporal commitment, and behavioral coherence—into reproducible, empirical indicators.

The goal is to create an operational standard for classifying systems based on the degree to which they act as if they possess intentional orientation. FIT supports improved auditing, documentation, and oversight by identifying when systems exhibit functional intentionality sufficient to trigger heightened scrutiny or liability.

Crucially, the framework rejects electronic personhood and does not treat AI as moral agents. Instead, FIT strengthens accountability by re-anchoring responsibility in the human and institutional actors who design, deploy, and benefit from these systems. Some forms of AI autonomy may be undesirable—especially in high-stakes settings such as healthcare, finance, or defense—where goal-directed behavior can magnify risks. FIT provides a clear, policy-ready tool for evaluating such systems without granting them legal or moral standing.