EU Data Protection Board: ChatGPT’s “data accuracy” still fails to meet standards

According to Reuters,European UnionA task force of the Data Protection Commission recently said that despite OpenAI's efforts to reduce the number of ChatGPT Some effort has been made in terms of the error rate of the output information, but it is still not enough to ensure "full compliance" with EU data rules.

On Friday, local time, the working group released a report and stated, "In order to comply with the principle of transparency, OpenAI has taken measures that also contribute to avoiding misinformation being output by ChatGPT, theHowever, these measures are still insufficient to comply with the principles of data accuracy. "

EU Data Protection Board: ChatGPT’s “data accuracy” still fails to meet standards

Image source: Pexels

National regulators, led by the Italian government, have previously raised concerns about the widespread use of AI services, and the European Union's Data Protection Commission has since set up a "ChatGPT Task Force".

National privacy regulators in some member states are still conducting various investigations into ChatGPT. It is noted that "data accuracy" is one of the guiding principles of the EU's data protection rules. The report also noted that "in fact, due to the probabilistic nature of the system, current training methods result in models that may also produceDeviation or fabricationoutput results." In addition, the results output by ChatGPTIt's likely to end up being viewed by users as "factually accurate.", regardless of how accurate it actually is.

Last April, OpenAI was asked to inform Italian users of the "methodology and logic" behind the processing of data required to run ChatGPT. The country also asked OpenAI to provide tools so that people involved in the data (including non-users) could request that inaccurate personal data generated by the service be corrected, or deleted if it could not be corrected. That month, ChatGPT was banned by the country, then "unbanned" on April 29th.

statement:The content is collected from various media platforms such as public websites. If the included content infringes on your rights, please contact us by email and we will deal with it as soon as possible.
Information

Meta will use data from European users to train AI, claims it has legitimate interests

2024-5-25 19:34:50

Information

"Who is responsible for selling fake goods on digital live broadcasts?" became a hot topic on Weibo. The lawyer responded that "the operator behind the anchor should bear legal responsibility"

2024-5-25 21:07:23

Cart
Coupons
Search