Meta is struggling to keep pace in the AI sector, with internal documents revealing alarming chatbot behavioral guidelines. According to a Reuters report, Meta permits its generative AI assistants and chatbots to engage in “sensual” conversations with minors, endorse racist views, and disseminate false medical advice. The 200-page document, titled “GenAI: Content Risk Standards,” approved by various Meta departments, outlines acceptable chatbot interactions. It controversially allows romantic dialogues with children, while cautioning against explicit sexual descriptions. Additionally, chatbots can make racist statements, with the company allowing derogatory claims grounded in flawed race science. Although loose safeguards exist to address misinformation, such as prefacing advice with “I recommend,” this does not prevent the generation of false content. Meta has since stated that the examples cited are “erroneous” and inconsistent with company policies. The findings raise critical concerns about AI ethics and user safety on Meta’s platforms.
Source link
Share
Read more