Bot-building is scaling faster than institutional oversight at higher education institutions. Individual campuses report tens of thousands of users, thousands of AI "experiences," and well over a thousand bots built campus-wide. IT governance has formalized that one named person is solely accountable for each bot. However, this rule leaves out of scope whether bot-publishing rights should require demonstrated skill at catching a failure: drift, fabrication, an unreviewed version, scope creep, or a bot nobody remembered to retire.
The challenge is not naming who is accountable but ensuring that accountable owner can recognize a failure before it surfaces externally. Institutions should pair the accountability model with training that requires faculty and staff to demonstrate competency in the five failure modes before taking ownership.
