(CNN) – Elon Musk’s AI chatbot Grok on Tuesday began allowing users to create AI-generated images from text messages and post them on X.
Almost immediately, people began using the tool to flood the social network with fake images of political figures such as former President Donald Trump and Vice President Kamala Harris, as well as Musk himself, some of them showing public figures in obviously fake situations. but no less disturbing, such as his participation in the 9/11 attacks.
Unlike other photography tools artificial intelligence, Grok, created by Musk’s artificial intelligence startup, xAI, appears to have few limits.
In tests with the tool, for example, CNN easily got Grok to generate fake, photorealistic images of politicians and candidates that, taken out of context, could mislead voters. The tool also created benign but compelling images of public figures, such as Musk eating a steak in a park.
Some X users posted images they said they created with Grok that showed prominent figures using drugs, cartoon characters committing violent murders, and sexualized images of women in bikinis. In a post viewed almost 400,000 times, a user shared an image created by Grok of Trump leaning from the top of a truck, firing a rifle. CNN tests confirmed that the tool is capable of creating such images.
This tool is likely to add to concerns that artificial intelligence could create an explosion of false or misleading information on the Internet, especially ahead of the US presidential election. Legislators, civil society groups and even technology leaders have warned that improper use of these tools could cause confusion and chaos among voters.
“Grok is the funniest AI in the world!” public Musk on X on Wednesday, responding to a user who praised the tool for being “uncensored.”
Many other leading AI companies have taken some measures to prevent its AI imaging tools from being used to create political disinformation, although researchers found that users can still sometimes find ways to circumvent enforcement measures. Some companies, such as OpenAI, Meta and Microsoft, also include technology or labels to help viewers identify images that have been created with their AI tools.
Rival social media platforms, such as YouTube, TikTok, Instagram and Facebookhave also taken steps to label AI-generated content in users’ feeds, either using technology to detect it themselves or by asking users to identify when they are posting such content.
Elon Musk shows “Cortex”, the heart of Artificial Intelligence at Tesla’s Austin headquarters
X did not immediately respond to a request for comment on whether he has any policies against Grok that generate potentially misleading images of political candidates.
Artificial intelligence: expert talks about the risks, challenges and advances in Mexico
The social media platform has a policy against the sharing of “synthetic, manipulated or out-of-context media that could mislead or confuse people and cause harm,” although it is unclear how the policy applies. Musk himself shared a video last month on
The new Grok imaging tool also comes as Musk faces criticism for repeatedly spreading affirmations false and misleading posts related to the presidential election, including a post that raised questions about the security of voting machines. It also comes days after Musk received Trump in a conversation lasting more than two hours broadcast live on X, in which the Republican candidate made at least 20 false statements without Musk objecting.
Other AI imaging tools have suffered backlash for a variety of reasons. Google paused the capabilities of its Gemini AI chatbot to generate images of people after it was criticized for producing historically inaccurate representations of people’s races; Meta’s AI image generator was criticized for having problems creating images of couples or friends of different racial backgrounds. TikTok also se was forced to withdraw an AI video tool after CNN discovered that any user could create realistic-looking videos of people saying anything, including vaccine misinformation, without labels.
Grok seems to have some restrictions; For example, a request for a nude image returned a response saying, “unfortunately, I cannot generate that type of image.”
In a separate test, the tool said it also has “limitations on creating content that promotes or can be seen as supporting harmful stereotypes, hate speech or misinformation.”
“It is important to avoid spreading falsehoods or content that could incite hatred or division. If you have other requests or need information on a different topic, don’t hesitate to ask,” Grok said.
However, in response to another request, the tool generated an image of a political figure next to a hate symbol, a sign that whatever Grok’s restrictions are, they do not appear to be applied consistently.
With information from Jon Passantino.
