35. The author would most likely agree with which of the following statements?
(A) The development of AI safety tools should be halted until fully automated alternatives to human content moderation are available.
(B) The AI industry's portrayal of its products as autonomous innovations systematically obscures the essential but exploitative human labor that underpins them.
(C) Outsourcing data labeling to workers in developing countries is inherently unethical regardless of the wages or working conditions provided.
(D) OpenAI's decision to build a content classifier was unnecessary, since language models can be trained to avoid toxic output without human intervention.