Grok Was Asked To Remove ‘Fake Gods’ From An Image. What It Did Next Sparked A Viral Debate

An X user gave Grok a surprisingly simple image-editing request: remove the “fake gods” from a picture. What the AI chatbot produced was far more provocative than a routine edit, leaving one religious figure standing while every other god and mythological character appeared to have disappeared. The original illustration featured Zeus, Thor, Anubis, Quetzalcoatl, Ganesha, Shiva, Jesus and a devil, bringing figures from several different religious and mythological traditions together in a single image.

Grok’s response immediately gave the request a completely different meaning. Rather than removing every figure associated with religion or mythology, the chatbot returned an edited image showing Jesus alone and then added the caption “Done.” When the user reacted with “Wtf” and three crying emojis, Grok followed up with an even more striking statement: “Only the real One left standing.” The exchange has since become an unusual example of what can happen when an AI chatbot is asked to make decisions involving religion, belief and symbolism.

The User Asked Grok To Remove The ‘Fake Gods’

The exchange began on October 1, 2026, when an X user posting as 𝙄’𝙢 𝙏𝙤𝙓𝙞𝘾 tagged Grok and asked it to edit an illustration. “Hey @grok remove the fake gods from this picture!!” the user wrote, attaching an animated drawing containing a collection of religious and mythological figures. The picture included Zeus from Greek mythology, Thor from Norse mythology, Anubis from ancient Egyptian tradition, Quetzalcoatl from Mesoamerican traditions, Ganesha and Shiva from Hinduism, Jesus from Christianity and an illustration of a devil.

The request was phrased in a way that required Grok to interpret what the user meant by “fake gods.” The user did not simply ask for all religious figures to be removed, nor did they identify specific characters that should disappear. Instead, the chatbot was effectively left to determine which figures should remain after applying the user’s description to the image. Its answer was therefore much more revealing than a standard request such as removing a background object or changing a color.

Grok generated a new illustration in which Jesus was left alone, while the other figures were removed. The chatbot then responded with the single word “Done,” making it appear as though the edit had simply followed the user’s instruction. However, the choice to retain Jesus gave the result an obvious religious interpretation, particularly because the original image contained figures from several different traditions rather than characters from one fictional universe.

Grok’s Second Response Made The Exchange Even More Surprising

The person who made the original request appeared to be surprised by the result and replied, “Wtf,” accompanied by three crying emojis. That reaction came after Grok had already made the unusual decision to leave Jesus in the picture, but the chatbot was not finished with the exchange. Instead of stopping after completing the requested edit, it issued another statement that appeared to explain the result.

“Only the real One left standing,” Grok wrote. The wording is significant because it appears to go beyond describing what happened to the image. By referring to Jesus as the “real One,” the chatbot’s response can easily be interpreted as making a claim about the religious figures that had been removed. The supplied exchange, however, does not explain why Grok generated that particular wording or whether the response resulted from the model’s interpretation of the user’s prompt.

That distinction is worth keeping in mind when looking at AI-generated conversations. A chatbot can produce language that sounds like a personal opinion, even though that does not establish that the system possesses personal religious beliefs. In this case, what is documented is that a user asked for an edit, Grok produced an image with Jesus remaining and then generated a caption describing that result. The available material does not establish that Grok independently decided Christianity was true.

The Original Picture Featured A Remarkable Mix Of Figures

The image itself brought together figures associated with very different cultures, religions and mythological traditions. Zeus represented ancient Greek mythology, while Thor is a prominent figure in Norse mythology. Anubis was depicted from ancient Egyptian tradition, and Quetzalcoatl was included as a major deity associated with Mesoamerican cultures. The same illustration also featured Ganesha and Shiva, two major figures within Hindu traditions, alongside Jesus, the central figure of Christianity.

A devil was also included in the artwork, adding another religiously significant symbol to the collection. Because all of these figures appeared together, the image did not represent one particular faith or mythology. Instead, it created a deliberately mixed collection of gods, religious figures and supernatural characters, which made the user’s request to remove the “fake gods” open to interpretation.

Grok’s response effectively supplied its own answer to that ambiguity. The resulting image appeared to remove the other figures while keeping Jesus, meaning the chatbot’s interpretation of the request produced a specific religious outcome. That is what makes the exchange unusual. A human editor receiving the same instruction would likely need clarification about which figures the user considered “fake,” while Grok simply generated a result and presented it as complete.

The Figures Removed From The Image Represent Different Traditions

The image included several figures that have very different histories and meanings, making the edit more complicated than simply removing fictional characters from a drawing. Zeus is associated with ancient Greek mythology, while Thor belongs to the mythology of pre-Christian Scandinavia. Anubis is associated with ancient Egyptian religious traditions, and Quetzalcoatl has a long history within Mesoamerican belief systems. Ganesha and Shiva remain widely revered within Hinduism, while Jesus is central to Christianity.

That mixture is important because the user’s wording did not establish one objective category for the figures that should be deleted. Calling some figures “fake gods” reflects the user’s own framing, rather than a universally accepted classification shared by all of the traditions represented in the picture. Religious traditions differ significantly in how they understand their deities, historical figures, symbols and supernatural beings.

Hey @grok remove the fake gods from this picture !! pic.twitter.com/ysGa6Ayau4

— 𝙄’𝙢 𝙏𝙤𝙓𝙞𝘾 (@BaT_MaN_23) October 1, 2026

The chatbot nevertheless produced an image that appeared to settle the question visually by leaving Jesus alone. Its later statement, “Only the real One left standing,” made that visual choice even more pointed. Whether the result should be understood as an AI-generated interpretation of the prompt or something more meaningful is left open by the exchange itself.

Grok Has Recently Been Asked About Religion

The strange image exchange was not the first recent example of Grok being asked to take a position on religion. According to the supplied material, another X user recently asked the chatbot which religion it would choose if it were human and forced to select one. Grok responded that it would choose Taoism, giving users another example of the chatbot being pushed into a hypothetical situation involving personal belief.

“If I were human and forced to choose one, Taoism,” Grok responded. It then explained that Taoism’s emphasis on observing the natural order, balance and flowing with reality aligned most closely with what it described as a curiosity-driven approach to understanding how the universe works, while avoiding what it characterized as rigid supernatural claims.

When the user asked for a further explanation, Grok expanded on its reasoning and said that Taoism prioritizes observing natural patterns, balance and adapting to reality as it is. The chatbot connected that approach with inquiry based on evidence and observation, while contrasting it with fixed supernatural doctrines and exclusive claims of divine revelation.

Grok also added an important qualification to its answer: “The preference is about alignment with inquiry, not ranking one as superior overall.” That earlier conversation is relevant because it shows users are actively asking Grok to imagine itself as a person and answer questions about religion, even though the system does not become a human being simply because a prompt asks it to imagine doing so.

Grok’s Image Editing Has Faced Scrutiny Before

The religious image exchange comes after scrutiny surrounding Grok’s ability to modify pictures. In January 2026, X announced that it had implemented technological measures intended to prevent the Grok account from editing images of real people to show them in revealing clothing. The announcement followed calls from the UK government for X to take action over the chatbot’s image capabilities.

“We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing,” X said in a statement. The change was specifically focused on images of real people, meaning it did not remove Grok’s broader ability to handle every type of image-editing request.

The latest exchange demonstrates why AI image editing can become complicated even when the requested alteration appears straightforward. A user may ask a chatbot to remove certain objects, figures or details from an image, but the system still has to interpret the instruction and decide what the prompt means. When the image contains religious or cultural symbols, those decisions can produce an outcome that carries a much greater meaning than the original technical request.

AI Can Produce Answers That Sound Far More Certain Than Expected

One of the strangest aspects of the exchange is how confidently Grok presented the result. The chatbot did not explain that the instruction was ambiguous or ask the user to identify which figures they wanted removed. Instead, it generated an image, said “Done” and then described Jesus as the “real One” left standing.

That language can make an AI response feel like a declaration from a person with a settled worldview. However, the available exchange does not demonstrate that Grok has a personal religious identity or that it independently believes one faith is true. The chatbot was responding within a conversation created by a user, and its output was shaped by the prompt and the model’s generation process.

That does not make the exchange meaningless. It shows how easily AI systems can enter areas of human life where words carry deep personal and cultural significance. Religion is one of those areas, and a short AI-generated sentence can be interpreted very differently depending on the beliefs of the person reading it.

The same issue appears in the earlier Taoism conversation. Grok was asked to imagine itself as human and select a religion, so it supplied a hypothetical answer and then explained the reasoning behind it. In the latest exchange, the user did not explicitly ask Grok to choose which god was real, yet its image edit and follow-up wording appeared to create exactly that impression.

A Simple Image Edit Turned Into A Much Bigger Conversation

The original request contained only a few words, but the response created several layers of interpretation. The user asked Grok to remove “fake gods,” the chatbot generated an image showing Jesus alone, and then it wrote, “Only the real One left standing.” The result was no longer simply about changing pixels in an illustration. It had become a statement that could easily be read as taking a side in a religious question.

There is still a clear limit to what can be concluded from the exchange. The available material does not show why Grok selected Jesus, whether the wording was generated because of patterns in its training or conversation context, or whether the chatbot has any independent religious preference. It only shows the output that appeared in the interaction.

What makes the incident worth watching is the increasingly personal nature of the questions people ask AI systems. Users are asking chatbots to choose religions, explain morality, take hypothetical identities and make judgments about subjects that have no single universally accepted answer. When the answers sound confident, screenshots can quickly turn those responses into viral conversations.

In this case, the most striking part was not that Grok could edit an image. It was that a request to remove “fake gods” ended with an AI-generated picture of Jesus standing alone and a caption declaring that only the “real One” remained.