The U.S. State Department used a map of Africa that mislabeled every country during a presentation at a global conference in Brazil [1].
The incident highlights the risks of relying on generative artificial intelligence for official government communications, particularly when accuracy in geography and international diplomacy is required.
The map was displayed July 30 [2] during the AIDS 2026 global health conference in Rio de Janeiro [1]. According to reports, 100% of the countries on the map were incorrectly labeled [1].
Analysis by the Reuters analysis team found that the image contained a watermark indicating it was generated by OpenAI's DALL·E [3]. The presence of the watermark suggests the department used an AI tool to create a visual aid rather than a verified cartographic source [3].
A State Department spokesperson said, "We made an unfortunate error" [4]. The department did not provide further details on how the image was vetted before being shown at the international event [4].
OpenAI, the creator of the DALL·E tool, is currently reviewing the situation. A spokesperson for the company said, "We are looking into the report" [5].
The use of AI-generated imagery in high-level diplomatic settings is rare, but this instance demonstrates how AI can produce "hallucinations"—confident but false information—that can lead to significant public embarrassment for government agencies [3].
“"We made an unfortunate error."”
This event underscores a growing tension between the efficiency of generative AI and the necessity of human oversight in government operations. By utilizing a tool known for visual inaccuracies in a diplomatic context, the State Department risked projecting a lack of attention to detail regarding African nations, potentially complicating the optics of U.S. engagement in the region.
