China, which has been trying to regulate both visual and text-based tools since the day artificial intelligence tools came into the agenda, often comes to the center of criticism as the regulations it wants shift more to the ‘restriction’ side. Of course, it would be a mistake to say that there is no group that wants the same regulations.
This time, chatbots like ChatGPT are in the focus of China. As we know, a huge amount of information flows to the artificial intelligence bots in question. China also wants to keep this flow under control, thus keeping the security and reliability of information under control. But there are also details you may not like.
This arrangement has two focuses: the information itself as the feed material and large language models (LLMs).
The proposal states that materials that feed artificial intelligence should not violate copyright and should not endanger personal information security. As a solution, it is recommended that the information to be given for training be given to artificial intelligence after being examined and approved by the relevant organizations and individuals.
Apart from this, LLMs such as Bard and ChatGPT are required to be built on models licensed and controlled by the authorities.
In addition, if more than 5 percent of information that is illegal or deemed harmful by cyber security laws is detected, it is requested that this information be blacklisted and prevented from being given to artificial intelligence.
What is meant by the illegal part here is things like violence, pornography, unfounded rumors and misinformation, and superstition. Of course, this is thought to include political information. There are concerns that this will hinder issues such as freedom of expression.
In the proposal, China states that developers should pay attention to these security provisions and that the information provided by users must undergo security checks. The proposal in question is awaiting feedback until October 25. What do you think the impact of this decision will be on China, which is currently on the agenda with censorship discussions?
14627
