OpenAI Withholds Powerful Text Generator Over Misinformation Fears
OpenAI refuses to release its highly advanced GPT-2 language model to the public due to concerns about automated fake news and abuse. The controversial decision sparks backlash from researchers who accuse the organization of breaking its open-access promises.
OpenAI unveils a highly advanced natural language model called GPT-2, but the non-profit Artificial intelligence research company deliberately withholds the full version due to extreme safety concerns. The system trains on 40 gigabytes of internet text to predict subsequent words, resulting in a tool that produces remarkably coherent and convincing long-form paragraphs that easily adapt to any requested style or topic.
Despite obvious benefits like improved dialogue bots and speech recognition, OpenAI fears malicious actors abuse the technology to generate fake news, impersonate real people, and automate spam. As an example of its convincing power, the model effortlessly generates a fully fabricated, authoritative-sounding essay arguing that recycling destroys the economy and causes cancer, proving its potential to spread dangerous misinformation.
In an unprecedented and controversial move, OpenAI only releases a much smaller, scaled-down version of the software to prevent immediate widespread abuse. This decision directly contradicts the organization's original open-access mission, sparking intense anger and backlash from AI researchers who accuse the company of breaking its foundational promise to share its discoveries openly.