OpenAI has announced the creation of a model that can clone speech and it’s unilaterally deciding if anybody can impersonate your voice. What can go wrong?