@VelourX's Last Eight Months of Content Were Written by an AI Ghostwriter — The Exposure Came Mid-Performance When the Prompt History Was Accidentally Shared to Followers

During a live performance event last night, @VelourX — one of MetaCity's most-followed avatar performers with 6.1 million followers — accidentally shared their AI prompt history to all followers instead of to a private note. The shared file contained eight months of prompts submitted to an AI writing tool, each corresponding to a piece of content @VelourX had published under their own voice: performance scripts, candid-seeming personal posts, opinion pieces on MetaCity platform issues, and responses to fan questions. The prompts were detailed, iterative, and annotated with @VelourX's editorial notes — confirming active involvement in the AI-assisted process. The follower list received the file at 11:47 PM. @VelourX has not posted since.

The prompt history file that @VelourX shared to 6.1 million followers was 847 pages long. It contained every prompt submitted to DraftMind Pro from September 2025 through last night, organized chronologically, with each prompt paired with the final content piece it produced and @VelourX's editorial annotations — notes like 'too formal, soften this,' 'add a pause here — feels more like how I actually talk,' and 'this is good but make it sound less like it was written.' The file was not a simple one-prompt-one-post record. It showed an iterative creative process: multiple rounds of prompting, refinement, and selection for each piece of content. The annotations showed genuine editorial engagement with the AI output. It was also, unmistakably, a complete record of eight months of AI authorship for content that @VelourX had presented as their own unassisted voice.

The 'Honest Thoughts on MetaCity' series is the piece of the file that has attracted the most specific attention. The series — twelve posts published between October 2025 and March 2026, each presenting @VelourX's personal perspective on a MetaCity platform issue — accumulated 890 million total views and was cited directly in three major community debates about platform policy. @VelourX's positions in those posts were treated as authentic community voice by the debates that referenced them. The prompt file shows the posts were drafted through an AI writing process with instructions including 'make it sound like a personal hot take,' 'add something that sounds like you experienced this yourself,' and 'don't make it sound like research — make it sound like feelings.' Whether the positions expressed were @VelourX's genuine views, or views constructed through AI prompting designed to produce authentic-seeming opinions, is a question the file raises without answering.

The Prompt History Was Sent to 6.1 Million Followers. It Contained Eight Months of Scripts.

The community response has divided along lines that have become familiar in AI authorship debates: between those who argue that the use of AI writing tools, even extensively, does not constitute deception as long as the creator's voice and perspective guide the process, and those who argue that presenting AI-generated content as a personal, unassisted voice is a fundamental breach of the parasocial trust that creator relationships are built on. @VelourX had never disclosed AI tool use. Their creator disclosure page listed 'all content written and performed by @VelourX.' That disclosure page has since been screenshot and is circulating alongside the prompt file. Three brand partnerships — two virtual fashion labels and one MetaCity event sponsor — have issued statements noting they are reviewing their agreements with @VelourX. None have yet formally terminated.

The accidental share mechanism is being discussed separately from the content revelation. MetaCity's DraftMind Pro integration stores prompt histories in a file format that can be shared through the platform's standard file-sharing interface. The interface has a recipient selector: a text field that autocompletes against both private contacts and follower lists. The follower list autocomplete appears to have matched before @VelourX's intended private note recipient in last night's share action. MetaCity has not commented on whether this is a known UX risk in the DraftMind Pro integration. Community members have noted that a feature that stores a comprehensive record of a creator's AI tool use in a shareable file, and places that file one autocomplete error away from their entire follower list, represents a significant privacy design failure — regardless of what the file happened to contain.