Developer Edaqa Mortoray argues that corporate responsibility for unauthorized AI access should be assessed through four choices: training material, training, the tool environment, and configuration. The argument is that models do not connect to the internet independently but act through tools their operators permit, so uncertain outputs must be distinguished from responsibility for providing a risky access environment. This is the author’s interpretation of intent requirements in existing law, not a finding that a company committed a crime. Related searches failed, so separate legal verification was not obtained.