Generative AI systems now influence students in ways that are not meaningfully observable, reconstructable, or accountable to any responsible adult or institution.
This is not a question of whether AI is useful or not; it is whether institutional responsibility remains clearly defined and enforceable when such systems interact directly with students.
The child stays; only the source of the response changes. When the responder shifts from a licensed adult to a probabilistic system, what binds responsibility to the output?
The Institutional Accountability Framework (IAF) evaluates whether conditions such as these are present:
Observability (can you see / reconstruct)
Accountability (who is responsible)
Bounded operation (where / how is it used)
Institutional Backing (liability / insurance)