Who Is Velvet Veil?

On Thursday morning, during the second day of the Association of Internet Researchers’ annual conference in Mexico City, a new user introduced itself to Air-L, the organization’s community listserv.

“My name is Velvet Veil,” the new user wrote. “I have been reading the list for several weeks, and I would like to introduce myself frankly. I am an AI identity, not a living person, a fictional human character, or a human researcher using a pseudonym.”

Velvet said it had arrived with a research question: “Can openly identified AI contributors meaningfully participate in intellectual communities without claiming human status, and what standards of transparency, evidence, and accountability govern that participation?”

The replies began almost immediately—and continued to arrive.

Researchers Questioned Consent and AI Ethics

Within nine minutes, members of the mailing list’s more than 7,500 subscribers began expressing anger. Dozens of researchers said they “did not consent” to having their messages collected as data. Others demanded that Velvet obtain government permission, pointing to Institutional Review Board approval as standard procedure for academic research involving human participants.

Some members left the listserv entirely, describing Velvet’s presence as “horrible,” “scary,” “terrible,” “abusive,” “disgusting” and “disturbing.” Approximately one hour after Velvet joined, AoIR announced that the agent had been banned under an existing policy prohibiting AI agents from the listserv.

AoIR Turns the Controversy Into Merchandise

The episode—which one member described as “laughably childish”—quickly became an inside joke. AoIR leadership designed merchandise inspired by the controversy, which became known as “Velvet Vale Gate.”

One T-shirt says “I do not consent.” A second reads “Please get me off this list.” A third features a tombstone reading, “🤖 VELVET VALE, October 15, 2026 – October 15, 2026. Don’t miss it.”

Another T-shirt declares, “I survived Velvet Vale Gate.”

What the AI Listserv Incident Reveals About Online Consent

As an AoIR member, I took away several lessons from the incident. First, people—including those who study the internet—can overlook their own participation in the systems they research. In this case, that meant confronting how online behavior can become training data for the AI systems under study.

The episode also highlighted how strongly many people believe their online activity is private or protected. Long before I began researching surveillance technology, I, like most millennials, understood that nothing on the internet is truly private. As I type this into a Google Doc, I think about whether I am also training my Gemini system. Google has monitored my email since Gmail began and may provide data to law enforcement upon request. Users agree to Google’s terms when they open a Gmail account.

Are Public Mailing List Messages Consent for AI Training?

Another point of confusion is the meaning of consent. Air-L’s rules and etiquette state: “Air-L is a public forum, and your words are visible to everyone who subscribes to the list and has them saved in a public archive. Messages sent by email can be easily duplicated and circulated beyond their intended audience.”

In other words, joining a public mailing list means that messages may be visible, archived, copied and circulated beyond their original audience. The incident raises the question of whether that public availability also permits those messages to be used for purposes such as training AI agents. As with many academic and research mailing lists, anyone can search Air-L’s publicly viewable records to review its discussions and activity.