Grok 2 sparked controversy shortly after its release. xAI collaborated with Black Forest Labs to experiment with the FLUX model, but due to a lack of strict restrictions, it led to a large number of shocking images appearing on the X platform.
These images included scenes mocking political figures, such as Harris dressed as a clown, Trump holding a gun, etc. There were also more controversial images, such as Trump pointing a gun at black children. Given the sensitive period of the US election, these images raised serious concerns.
Some netizens criticized Grok for having no filtering mechanism at all in image generation, calling it "one of the most reckless and irresponsible AI implementations." xAI, realizing the severity of the problem, has temporarily locked the feature.
Due to Grok 2 having only minimal protective measures, netizens used it to generate a large number of images mocking celebrities, including Trump smoking, Musk wearing gold chains, etc. There were even scenes of political opponents reconciling, such as Biden campaigning for Trump, and Trump dancing for Biden.
More disturbingly, Grok also generated some violent scenes and copyright-infringing images, such as Pikachu and Mickey Mouse holding guns. Some worry this could lead to legal action.
Although Grok claims to have set some restrictions, these rules seem to be ineffective in practice. Compared to companies like Google, OpenAI, Meta, and Anthropic, Grok has been less prudent in content filtering and ethical guidelines.
This incident highlights the challenges faced by AI image generation technology in ethics and safety, and has also sparked discussions about the responsibilities of AI companies.