Meta and OpenAI sued again for copyright infringement
Date: 2023-07-11 Author: Karina Ziganova Categories: BUSINESS
Comedian and author from the United States Sarah Silverman, along with writers Christopher Golden and Richard Kadri, are suing the developer of the ChatGPT chatbot - OpenAI - and Meta Corporation (recognized as extremist and banned in the Russian Federation).
"In the texts, some of the data is incorrect, but it is still clear that ChatGPT uses the content of specific works in the training dataset," the lawsuit says.
The plaintiffs allege that artificial intelligence (AI) models were trained on their jobs. The authors' works were obtained from pirate sites that "have long been used to train AI."
According to the authors, they did not consent to the use of their copyrighted books as educational material for ChatGPT. Meta developers used the authors' works to train LLaMA (Large Language Model Meta AI) language models.
Another OpenAI lawsuit
In late June, law firm Clarkson filed a class-action lawsuit against OpenAI for identity theft. In the 15-point lawsuit, the lawyers pointed to breach of confidentiality, negligence due to failure to protect personal data, and theft by illegally obtaining large amounts of personal data to train their models.
Moreover, OpenAI used the personal data of hundreds of millions of social media users. Thus, the developer got 300 billion words to train the chatbot. Moreover, the company "did it secretly and without registering as a data provider, as required by applicable law."
ChatGPT accounts leaked to the darknet
Prior to that, OpenAI got into another scandal — Group-IB experts found 101,134 ChatGPT accounts in the logs that cybercriminals put up for sale on the darknet. Analysts claim that the data was stolen with the help of viruses that hunt for the victim's information on personal computers.
One of the causes of data leakage is the use of a chatbot at work. The chatbot stored a history of user requests and responses, so the attackers who stole the logs can see all the victim's correspondence with ChatGPT.
