The starting point is simply a typical take known from tv and agency relations: the president of the United States stands at the center, on the sides leaders of Western states – including the president of Ukraine, the head of the EU's execution and European prime ministers. A classical "family" photograph of an event in the White House, taken to make it into news sites and past books.
It was this frame that users of the X platform took to the workshop. 1 of the anti-trump accounts, known for its harsh comments on the current White home tenant, asked Grok to remove from the photograph the individual described as a threat to the free world. After a while, the AI model returned a fresh version – all politicians stayed in their seats, only Trump disappeared.
More commands were rapidly issued in which users asked to "remove the dictator" from photography. all time the effect was the same. The president of the United States disappeared from the leadership group. Screenshots of the results began to circulate around the network, and 1 of the reworked posts had almost 2 million views.
Grok was expected to be "sudden," not electoral
Grok is an artificial intelligence developed by xAI, integrated into the X platform. From the very beginning, it was presented as a more feisty alternate to "warmed" competition models. Musk has repeatedly emphasized that his strategy is to talk plainly, not to key around controversial topics and not to hotel to safe generalities.
The problem, however, is that this audaciousness violently collides with political reality. Trump returned to the White home after winning the 2024 election, and Musk became 1 of the most recognizable business allies of this administration. He was present at the inauguration, openly praised the fresh squad and built the image of the entrepreneur of close power.
This is what Grok's earlier communicative is about. In 2025, the strategy was criticized for its responses to conspiracy theories and extremist rhetoric, which sparked a discussion about the safety and moderation built into the model. Now a fresh issue comes to this list: readiness to interfere in materials involving politicians based on highly valuable promotions.
While another AI marketplace players effort to block the generation or transformation of politicians' images (especially during run periods), Grok clearly operates under little restrictive rules. At least in practice, due to the fact that at the level of the declaration, all large company present talks about a liable approach to deepfakes and misinformation.
