xAI’s AI chatbot Grok-2 was launched on Tuesday that lets users to create AI-generated images from text prompts and post them to X. This image generation tool is quite similar to Google’s Gemini or OpenAI’s DALL-E but with some dissimilarities. It puts far fewer restrictions on the types of images that can be created. Almost immediately, people started employing this AI assistant to flood the social media site with fake pictures of political figures such as former President Trump and VP Kamala Harris, as well as of Elon Musk himself — some showing the public figures in clearly false but disconcerting situations, like participating in the 9/11 attacks.

xAI created Grok-2 seems to have fewer guardrails compared to other mainstream AI photo tools created ostensibly by competitors of xAI. In tests of this AI assistant, for instance, CNN was easily able to get it to generate fake, realistic pics of politicians and political contestants that, taken out of context, could be deceptive for voters. Grok-2 also created benign yet convincing pictures of public figures, such as Musk having steak in a park.

Some X users posted pics they said they created using Grok showing notables consuming drugs, cartoon characters committing violent homicides and sexualized pics of women in bikinis. In one post viewed almost 400,000 times, a user shared an image created using Grok of Donald Trump leaning out of the top of a truck, firing a rifle. CNN tests confirmed the AI assistant is capable of generating such images.

The AI tool is likely to add to apprehensions that artificial intelligence could cause an explosion of false or misleading info across the internet, particularly ahead of the US presidential election. Civil society groups, lawmakers and even tech leaders themselves have issued warnings that the misuse of such tools could cause confusion and chaos for voters. But Musk wholeheartedly welcomed the arrival of Grok-2. “Grok is the most fun AI in the world!” Musk posted on X on Wednesday, in reaction to a user praising the AI assistant for being “uncensored.”

Many competitors of xAI have taken some steps to avoid the eventuality of their AI image generation tools being used to create political misinformation, but researchers found out that users can still sometimes find ways around enforcement measures. Some companies, including Meta, OpenAI and Microsoft, also include technology or labels to empower viewers to identify images that have been made using their AI tools.