How can enterprises protect against the unique vulnerabilities of AI agents? Consider treating them as their own identities.