(The Atlantic) – Even as they claim the right to train their models on work belonging to other people, the AI companies have rejected similar reasoning when it comes to their own products. Consider OpenAI’s terms of service for ChatGPT, which forbid use of the bot’s “output to develop models that compete with OpenAI.” Anthropic, Google, and xAI have similar clauses forbidding people from using the material generated by their chatbots to train competing products. In other words: We can train on your work, but you can’t train on ours.

In the current economic environment, it’s not surprising that companies vying for market dominance would operate with standards that serve their bottom line. But it’s striking nonetheless how sharply their actions can contradict their professed values. (Read More)