It’s reported that Open AI discussed with US government officials about technology theft of China’s Deep Chic. A number of the hypocrisy identified that ‘web crawling’ and ‘distillation’ are different.
Chris Rihan Open AI chief global business manager said in an interview with Bloomberg TV on the tenth (local time) about Deep Chic’s open AI technology theft, “Now we have confirmed some evidence and continues to review. I discussed about it. ”
Deep Chic surprised the technology industry by launching a high -performance AI model at low price last month.
Nevertheless, Microsoft’s security researchers informed Open for the primary time that Deep Chic is inappropriate data through the API of Open AI. Open AI said that Deep Chic is considering and considering the signs of coaching the model from the info that’s inappropriate from their models by distillation.
Nevertheless, some criticize this response to open AI as hypocrisy. It’s identified that open AI, which collects copyright contents and used for AI model training, used its AI model output for training.
But Lihan insisted that there was a transparent difference on this. He compared the training approach to open AI to “loaning books within the library.”
But Deep Chic explained, “It’s near bringing a library book and changing the duvet and the writer.”
“It is a cloning, and we’re concerned,” he said.
In actual fact, Open AI stated that copying data through model terms is a violation of the foundations.
Nevertheless, experts say that if this problem goes to court, it’s difficult for open AI to win. It’s analyzed that using ‘process use’ that Open AI insisted in other lawsuits may hold onto his ankle.
By Park Chan, reporter cpark@aitimes.com
