A flood of questionable and sometimes offensive images has been seen by many users on the platform X over the past hours, following the release of the latest version of the chat robot Grok, created by billionaire Elon Musk. Since yesterday, this newest version of the Grok chatbot has appeared for the first time, equipped with a new image generation tool. However, the dangers of this version quickly surfaced, as many AI-generated images emerged.
These strange and offensive images depicted political figures and celebrities in compromising situations. For example, images circulated showing former U.S. President Donald Trump piloting a plane toward the World Trade Center, as well as images of global singer Taylor Swift and U.S. Democratic presidential candidate Kamala Harris either nude or in their underwear. Unlike OpenAI's ChatGPT, Grok can produce a variety of images that similar AI tools have banned due to violation of rules regarding misinformation and misuse. For instance, ChatGPT cannot generate images of an actress or celebrity nude, or of an activist or political official in their underwear, as reported by the British newspaper The Guardian.
Furthermore, Grok's image generator does not refuse any requests that involve a breach of copyright and intellectual property rights, unlike most other AI tools, including ChatGPT. Grok has, for instance, generated images of Mickey Mouse greeting Adolf Hitler or Donald Duck using heroin. This comes as many Western countries are striving to establish laws to limit the proliferation and unchecked growth of AI tools and programs amid concerns about their impact on the spread of information and manipulated images, which can sometimes affect security situations and even elections, both in the United States and other countries.