Khaberni - Media reports revealed a leak of content from the Grokipedia "conservative" encyclopedia, which was created by artificial intelligence and developed by Elon Musk's company xAI, into the responses of ChatGPT; this raised many questions about the accuracy of the sources these systems rely on.
The Guardian reported that the GPT-5.2 system cited Grokipedia nine times in response to more than a dozen different questions, indicating that the content of Musk’s encyclopedia is beginning to exceed the controls within his own system.
xAI launched the Grokipedia website last October, following Musk's complaints about Wikipedia's bias against conservatives. However, the encyclopedia faced sharp criticism after direct articles from Wikipedia were detected, including "ideological justifications" for slavery, the use of derogatory terms, and linking global health crises to pornographic material.
The Guardian noted that ChatGPT did not cite Grokipedia in major cases, such as the Capitol attack on January 6 or the AIDS epidemic, but it was cited in lesser-known topics, including claims about public figures that had been debunked.
The reports also indicated that Robert Clood from the Anthropic company also used Grokipedia to answer some inquiries.
In an official response to these developments, a spokesperson for OpenAI told The Guardian that the company "aims to benefit from a wide range of sources and views available to the public", without providing further details about the filtering criteria for controversial content coming from xAI platforms.
