LONDON, UNITED KINGDOM — We now all know the cursory signs of AI-generated visual content, from the stubborn desire to tint everything yellow to the trend of everything looking faux-gilded-polished. Now, a cursory search indicates that minority representative groups and political claimants, ranging from a self-proclaimed administration for the Chagos Archipelago to the statutory council of Serbia’s Vlach ethnic minority, have formally adopted institutional emblems generated through commercial artificial intelligence models. The adoptions certainly mark a tangible departure from traditional civic heraldry, where formal heraldic grants and vexillography is replaced with raw outputs produced by LLMs.

The self-declared Government of the Chagos Islands, established in London in December 2025 by British-Chagossian veteran and activist Misley Mandarin to contest the transfer of the British Indian Ocean Territory to Mauritius, promulgated an AI-generated seal as its primary executive insignia.

Seal of the Government of the Chagos Islands

Created to accompany the unrecognized body’s legal petitions in the British Indian Ocean Territory Supreme Court and its physical encampment on Île du Coin in the Peros Banhos atoll, the work synthesizes traditional British imperial heraldry with regional Indian Ocean iconography, displaying twin leaves, stylized stars, and an imperial crown. Due to its nature as an unrecognized entity, it lacks a formal blazon issued by the College of Arms or a statutory royal patent, both of which are standard prerequisites for British Overseas Territory insignia.

Meanwhile, in Eastern Serbia, the National Council of the Vlach National Minority enacted a similar measure during a formal session in Požarevac, adopting a new flag and coat of arms to represent the community in the Timok Valley, Braničevo, and Bor districts. The council’s new achievement, centered on an azure shield charged with a white double-headed eagle within a narrow red border, alongside a horizontal multicolored flag featuring a nine-rayed sun set over mountain peaks, was confirmed through an administrative decree and subsequently uploaded to the council’s national registry. The publicly available source rendering is quite literally an unedited 1024-by-1024 pixel bitmap, exhibiting line blending and structural asymmetries consistent with generative machine learning tools. The move distinguished the Eastern Serbian assembly from minority organs in the Autonomous Province of Vojvodina, where the distinct Romanian National Council maintains separate, traditionally drafted heraldic symbols.

The AI-generated flag and coat of arms, respectively.

The procedural shortcutting may draw scrutiny across vexillological registries and intellectual property circles, primarily due to the non-conformity of AI files with classical rules of tincture and reproducibility.

ZeroGPT, an AI-Generated asset detector, classifies the image as created by AI.

International heraldic conventions stipulate that institutional arms exist fundamentally as a written blazon from which any heraldic draftsman can produce an authorized physical emblazonment, and by contrast, algorithmically compiled emblems function solely as static raster images with embedded artifacts that complicate physical embroidery and vector reproduction. Not to say the least that in the case where such symbols have been infused with watermarks, the adopted symbols now is adorned with a literal symbol of a multinational AI company stuck into its fabric, which may be a point of concern for such drafters.

To make matters worse, under prevailing copyright doctrines maintained by intellectual property offices in the United Kingdom, the European Union, and the United States, computer-generated visual outputs lacking demonstrable human creative input remain ineligible for copyright protection, leaving the official administrative markers of both bodies legally unowned and in the public domain from the moment of their annoucement. Despite this, the sheer ease of making flags and other heraldry this way hints that these incidents may not be isolated as the years go on.