< IMG LOADING = "Lazy" SRSC = "/Sites/Default/Files/Styles/Medium/2025-02/Stefani%20REYNOLDS%20AFP.JPG ? Itok = MZO6EDID" Width = "1300" Height = "731" alt = "open ai" Class = "Lazyload Img-Fluid Image-Style-Max-1300x1300" SRC = "/Sites/Default/Files/Styles/Max_1300x1300/Public/2025-02/Stefani%20REYNOLDS%20AFP.JPG ? Itok = JY7IMW"/> Reynold/AFP < P > Last Friday, the American company OPENAI, editor of generative artificial intelligence Chatgpt, has announced that it has banned user accounts in China and North Korea for “malicious” use of its model. The decision, made public in a security report, evokes disinformation and surveillance campaigns targeting Facebook, X or YouTube, Telegram and Reddit social networks. Its chatgpt model was also used to create a tool for these same purposes.
< P >OPENAI explained that it had detected several abusive use cases of its technologies. In its report, the emphasis is put on an operation baptized by the company “Peer Review”, led by Chinese actors, to refer to the language used as well as the activity hours. “We have banned a group of accounts Chatgpt which, on the basis of behavioral patterns and other elements, probably came from China, ”it is said in the Report. < P >< Strong > Influence, surveillance and fraud campaigns < P > “These accounts used our models to analyze documents, generate sales arguments and descriptions Tools for monitoring activities on social networks fueled by third -party models, editing and debugging the code for these tools, as well as looking for political actors and subjects ”, continues Openai. < P >Concretely, “Peer Review” aimed to develop a social media surveillance tool, named “Qianyue Overseas Public Opinion AI AI AI AI AI”, having to analyze in real time publications and comments on various platforms like X, Facebook, Youtube, Instagram, Telegram and Reddit. The goal, to refer to the same source, was to collect information on conversations relating to Chinese political subjects, emphasizing the calls to be demonstrated, and to transmit this data to the Chinese authorities as well as to its Embassy abroad. & nbsp;< P >The United States is not the only watched country. Germany and the United Kingdom also appear in the short-list of the targets of these networks. As for the most monitored themes are the question of Uighurs or diplomacy in the Indo-Pacific region, backyard of Beijing in a context of strong tensions with Taiwan and India, to name only these two countries. & Amp ; nbsp;
< P > networks also exploited Chatgpt for purposes of Disinformation, adds Openai, evoking in his report another case involving the generation of press articles in Spanish criticizing the United States, which was then published as sponsored content in Latin American media under the signing of a company Chinese. This operation, baptized “sponsored discontent” by Openai, would be a first according to the American company, which specifies that no campaign of influence of Chinese origin has previously managed to place long articles in the Latin American media to criticize The United States.
< P >Openai reveals that another disinformation campaign has been carried out and this time, they were not press articles but posts on social networks. Her chatbot was used to generate short messages published on social networks, notably denigrating the publications on X Cai Xia, former professor at the Central School of the Chinese Communist Party (PCC), dissident in the eyes of Beijing. & Amp; nbsp; < P >< Strong > of the diagrams Advanced < P >North of this superpower, that is to say in North Korea, other activities have been detected. It is mainly a question of employment fraud targeting Western societies. A well -cut and already documented system, which consists in pretending to be Western employees by having a false identity, including on the Internet, with a triple objective, divert funds through accumulated wages, obtain equipment like professional computers as well as information on the employer.< P > The American company thus banned the accounts involved, detected at The help of several of his IA tools. In her report, published regularly to “prevent the use of his technologies by authoritarian regimes in order to strengthen their power or their control over their citizens”, Openai says that it is the first time that she has observed an IA surveillance system also advanced.
< P > It is not the first case of “diversion” of Chatgpt according to Ben Nimmo, principal investigator at The American startup, which had mentioned, at the summit in Paris in early February, similar cases thwarted by its teams. & nbsp;< P >This ban is part of a broader context of AI racing between the major world powers. The United States, China and the European Union are in common in the development of AI, seeking to dominate an area that deeply intersects with global power dynamics, data confidentiality and national security concerns.
< P > At the beginning of the year, the Chinese Deepseek was launched by a Chinese company from IA, Destabilizing for a few days the scholarship lessons in the West. In the process, several Western countries, including American states, Italy and the Netherlands, have prohibited the use of this AI, in particular their senior officials, justifying their decision by the risks of national security and protection of the protection of data.< P >< Strong > The article you liked ? It mobilized our editorial staff which only lives from your gifts. < Br > Information has a cost, especially since competition from subsidized writing requires additional rigor and professionalism.
< P > With your support, France-Soir will continue to offer its articles free of charge & nbsp; we think that everyone must have access to free and independent information to forge their own opinion. < /p > < P > You are the sine qua non condition for our existence, support us for France-Soir to remain the French media which makes the most legitimate express.