Kathrine Frich
Artificial intelligence tools are increasingly shaping the information landscape worldwide.
While companies promote these systems as tools for research, creativity and communication, they are also being monitored for potential misuse.
A new report from OpenAI highlights how some actors have attempted to use AI systems to generate large volumes of political content online.
Alleged propaganda networks
OpenAI said in a report titled Disrupting Malicious Uses of Our Models that Kremlin-linked actors used ChatGPT and other tools to produce large amounts of content for influence campaigns.
According to the report, one of the operations involved the Russian network Rybar, which runs a large Telegram channel and maintains accounts on other platforms.
OpenAI researchers said users generated articles, social media posts and comments with the help of ChatGPT. Some content was written in English and Spanish even though prompts were submitted in Russian.
The campaign’s narratives included praise for Russia and its allies, criticism of Ukraine and accusations of Western interference.
Influence campaigns
OpenAI said the operation produced posts targeting both Russian and international audiences.
One example cited in the report involved an article accusing Germany of building an influence network in Moldova. In another case, users asked ChatGPT to generate comments to accompany the article.
The report also described a separate proposal for election-related activities in several African countries.
These plans reportedly included both online messaging campaigns and offline activities such as building networks of local supporters and organizing public events.
Activity in africa
Another operation identified in the report involved accounts generating long-form geopolitical content focused on sub-Saharan Africa.
Some of the material appeared on Facebook pages posing as news outlets in countries including South Africa, Ghana, Kenya and Angola.
The articles often praised Russia while criticizing Ukraine, the United States and the United Kingdom.
According to OpenAI, some posts also targeted individual leaders including Ukrainian President Volodymyr Zelenskyy and U.S. President Donald Trump.
Fake identities
The report said many of the articles were published under the names of fictional authors.
OpenAI researchers said they checked each listed writer and found no evidence that the individuals existed.
The company said it blocked the accounts involved in the activity after identifying the operations.
Sources: OpenAI, United24Media.
